THE LOWER OF KANSAS. PART 9. THE ORDERS , RAPHIDIODEA, CALONEURODEA AND (PROBNISIDAE), WITH ADDI- TIONAL PROTODONATA AND ."

Received October 12, 1941 Presonted October 9. 1941 Dr. Tillyard's sudden deatli in 1938 left unfinished new wing is 195 min. long and 40 mln. wide (inaxi- his account of the Permian insects from the Elmo murn). This indicates that the total wing expanse limestone in the Peabody Museum at Yale Uni- was about 41 cm. or 16 inches. The holotype speci- rersity. The unstudied specimens in that collection, men, which was not quite complete, measured 165 mostly belonging to the Protorthoptera and related mm. in length.' groups, have recently been turned over to me by Tillyard's restoration of the wing of schucherti the authorities of the Peabody Museum for identifi- (1925, fig. 5) is now seen to be incorrect in two re- cation and description. The present paper deals spects. First, the wing is more slender apically than with the Caloneurodea contained in both the Yale he assumed it to be; and second, tlie subcosta ends and Harvard Collections, and with the Neuroptera, well before the apex of the wing, instead of at the Raphidiodea, and part of the Protorthoptera (family apex, as lie lias drawn it. His figure of the actual Probnisidae) in the Harvard' collection, the .Yale type specimen is incorrect in that it has lt4 + 5 specimens of these groups having already been de- (termed R3 in his paper) arising in the distal half of scribed by Tillyard. I have also included here an the wing, well beyond the level of the first fork of account of some Protodonata and Megasecoptera in MA. Actually, R4 + 5 arises (in both the holotype the Harvard collection which were not considered in and the new specimen) slightly before the middle of my earlier papers on these orders. the wing, and directly above the fork of RIB. In all except individual details tlie new wing has tlic saine Additional PROTODONATA venation as the holotype. Since the publication of my previous account of One of the most interesting aspects of the new the Protodonata from the Elmo limestone (1939), is the presence of an oblique vein at the point Professor R. C. Smith, of Kansas State College, has of separation of R2 from R3. This is obviously the sent me a magnificently preserved hind wing of homologue of the vein (subnodus) described by Megatypus schucherti Tillyard (family Meganeu- Sellards in Typus permianus (1906, p. 253, figs. 4, 5). ridae). It was collected by two entomological As I have already stated (1939, p. 38) such a vein students in the College, Mr. Otto Wenger and Mr. has not previously been discernible in any Proto- Floyd Holmes; the reverse of the fossil is in the donata (Megatypus or Typus) in the Yale or Har- Kansas State College entomological collection and vard collections from Elmo. It should be noted the obverse in tlie Museum of Coinparative Zoology. that there is no indication of the oblique vein "oJ' Apart from doubtful fragments, M. schucherti lias described by Sellards in Typus. The obliquity of previously been known only from the holotype, a this vein is almost certainly an individual trait. hind wing lacking the distal fourth of the wing and some details of venation (Tillvard, 1925). The new Additional AIEGASECOPTERA bpecimen is an absolutely complete hind wing, The extinct order Afegasccoptera is not uncoin- shoa,ing with perfect clarity all veins (Plate I, inonly represented in certain strata figure 1). It is not only interesting as the largest of Europe, but until comparatively recently it has entire wing yet found in the Permian, but it is also not been found in Permian rocks. The insects which i~i~portantbecause it enables us to complete our Tillpard originally described from the Elmo lime- howledge of the venation of this huge . The stone as members of a new order, Protohymenoptera, * This investigatiorl has been aided by a grant from The width of th~holotype wing was given by Till- tile Permanent Science Fund of the American Acade~rly yard as 94 mm., but this is an error; it sl~ouldhave of Arts and Srienres. been 39 mIn. are now generally recognized as close relatives of the Bltt~ouTillyarcl Megasecoptera and as constituting merely a sub- Elmoa Tillyard, 1937, Amer. Journ..Sci. 33: KX. order of the Megasecoptera. In his last discussion Fore wing: long and moderately slender; costal of the affinities of the Protohymenoptera, Tillyard space of moderate width; Sc terminating on Rl (1936) expressed agreement with this view, though before mid-wing ;Rl unbranched; pterostigma absent claiming it to be a matter of personal choice whether or at most very weakly formed; Rs with at least the group be regarded as a suborder or an order. three terminal branches; R, M and CuA separating The Protohymenoptera from the Elmo limestone at the same point; MA probably unbranched; MA contained in the Harvard collection have already joined to Rs by a weak cross-vein shortly after the been described (Carpenter, 1931a, 1931b, 1933, origin of the latter; MP forked; CuA unbranched; 1939). cross-veins few and weakly developed. Hind wing: Apart from the Protohymenoptera, the order similar to the fore wing in venation, except for dif- Megasecoptera is represented in the limestone by ferences noted under the account of the family. several species which are more closely related to the Genotype: Elmon trisecta Till. Eumegasecoptera2 than to the Protohymenoptera. One of these, Elmoa trisecta Tillyard, was recognized Elmoa trisecta Tillyard by Tillyard as a Megasecopteron and placed by him in a new family, Elmoidae. Two other species which Figure 1 he described from the formation (Martynovia insig- Elmoa ~risecta Tillyard, 1937, Amer. Journ. Sci. nis and Martynoviella protohymenoides) are, in my 33: 84. opinion, also Megasecoptera, though they were con- Fore wing: length, 12 mm. ; width, 3.2 mm. ; 3 or 4 sidered by him to be Neuroptera, belonging to the strong costal veinlets at base, above stem R+M, suborder Sialoidea (). Unfortunately, and 1-3 weaker ones more distal along Sc; a few not only was each of these three species based upon oblique veinlets between R1 and costal margin, a single wing, but the type of Elmoa trisecta con- especially distally; Sc terminating beyond origin of sisted of a basal fragment of a wing. Had these Rs; stem R + M arched; R4 + 5 long, arising before specimens been more nearly complete, Tillyard mid-point of Rs; R2 and R3 at most half as long as would undoubtedly have recognized the very close R4 + 5; MA arising proximally to origin of Rs; MP relationship between Elmoa, which he placed in the forked to about half its length; CUPnot forked, but Megasecoptera, and the Martynoviidae, which he connected to hind margin by a long cross-vein; 1A assigned to the Xeuroptera. The evidence for my remote from hind margin, connected to it by at conclusion on the affinities of these is derived least one cross-vein; 2A very short. from a study of the specimens in the Harvard col- Hind wing: length, 12 mm.; width, 4 mm.; more lection and will be considered below, after a descrip- nearly oval than the fore wing; costal space as wide tion of the insects. as in fore wing, with similar veinlets; Sc not quite so long as in fore wing and nearer R + M than in fore Suborder EUMEG.~SECOPTERA wing. Family ELMOIDAE The body structure is preserved in one specimen (No. 4591ab); this shows only the thorax, part of Wings subequal; membrane glassy. Fore wing: the head, and a small piece of the abdomen. The Sc terminating before or at mid-wing; costal veinlets head is small (1.5 mm. long) and broad, and appar- few and weakly developed; Rs arising before mid- ently has prominent eyes. wing; M coalesced proximally with R; MA and MP Holotype: No. 15591ab, Peabody Museum, Yale present; MA not coalesced with Rs or R1; CuA University. This consists of the basal third of a coalesced proximally with stem R + M; two anals hind wing, not a fore wing, as stated by Tillyard. present ; pterostigma apparently absent. Hind In the Harvard collection there are eight speci- wing: narrow basally, but broader near the middle mens of trisecta, of which six are from the upper than the fore wing; Sc closer to R than in fore layer of the limestone: no. 4591ab, a complete insect, wing; rest of venation as in fore wing. Almost showing both pairs of wings; the fore wings overlap nothing is known of the body structure. part of the hind pair, but the two pairs are readily At present the family consists of one genus. distinguished. No. 4606ab, a very nearly complete The order Megasecoptera includes the two sub- fore wing, well preserved; no. 4595ab, 4607ab, each orders Eumegasecoptera and Pr~t~ohymenoptera(Car- consisting of the proximal third of a fore wing; no. penter, 1933). 4590ab, a complete and excellently preserved hind THE LOWER PERMIAN INSECTS OF KANSAS. PART 9

FIGURE1. Elm.oa trisecta Till. A, fore wing: drawing based mainly on specimen no. 4604, Mus. Comp. Zool. B, hind wing; drawing based chiefly on specimen no. 4590, Mus. Comp. Zool. Sc, subcosta (concave); R1, radius (convex); Rs, radial sector (concave); MA, anterior media (convex); MP, posterior media (concave); CuA, anterior cubitus (convex); CUP, posterior cubitue (concave); IA, 2A, anal veins.

wing; and nos. 4592ab, 4593, 4594ab, all hind wings, which it represents was more generalized than other more or less complete. These specimens were col- known Megasecoptera. As Tillyard has proposed, lected by F. M. Carpenter at the Harvard quarries the Elmoidae were probably Permian derivitives of at Elmo. a poup of Megasecoptera which must have existed Tillyard's figure of the base of the hind wing of well back in the Pennsylvanian Period. The short this species (1937, fig. 1) is incorrect in two impor- subcosta of Elmoa, however, is a specialization not tant respects. The subcosta terminates on R1 just previously found in any Eumegasecopteron, though beyond the origin of Rs, instead of continuing for a a similar subcosta occurs in the Protohymenoptera. considerable distance beyond this point, as he has Of particular interest in connection with the drawn it. The wings of nearly all specimens of affinities of Elmoa is the difference in the shape of trisecta are wrinkled (as in the Protohymenoptera of the fore and hind wings. The complete specimen in the Elmo limestone) and apparently one of these the Harvard collection shows conclusively that the wrinkles was confused by him with the end of Sc. broader wing is the hind one. This is consistent He has also drawn MA as forked shortly beyond its with the condition which occurs frequently through- origin; but this fork does not show in the type out the Megasecoptera as a whole, including such specimen or in any of the fossils in the Museum of slender winged types as Foriria maculata Meunier, Comparative Zoology. In none of the specimens of from the Carboniferous of Commentry. trisecta which have been found is MA preserved to So far as the Permian is concerned the following the actual wing margin, but the missing part is so family, Martynoviidae, seems to be the one most short that it seems almost certain that this vein was closely related to the Elmoidae. unbranched. Family MARTYNOVIIDAE Wings subequal; membrane glassy. Fore wing: Although Elmoa possesses a number of charac- Sc terminating on R1 at about mid-wing; costal teristics foreign to the Carboniferous and most veinlets few in number; pterostigma weak, but ex- Permian Megasecoptera, I believe with Tillyard tensive, the area between R and the costal margin that it belongs to this order. All of the Eumegase- beyond the end of Sc being slightly thickened and coptera previously known have possessed a petiolate pigmented; Rs arising before mid-wing; M coalesced or subpetiolate wing. The relatively broad wing basally with R; MA and MP present, MA coalesced base of Elmon therefore suggests that the family with Rs; C11 coalesced with R f M basally; two or 58 CARPENTER three anal veins present. Hind wing: much broader above the origin of Rs; Rl close to the costal margin than the fore wing, costal space slightly broader distally, unbranched; Rs with from 3-5 terminal than that of fore wing; venation like that of the fore branches; M separating from stem R + M at about wing, except that the pterostigma is absent or very the point of divergence of CuA from R + M; MA, weakly formed at most. The body structure is RtP, CuA and anals unbranched; cross-veins few in unknown. number. Hind wing: similar to the fore wing, This family was originally established by Tillyard except for differences noted under the family. (1932) for two genera, Martynovia and Martynovi- Genotype : Martynovia insignis Till. ella. In a later paper (1937) he also placed here the At present two species of this genus are known, genus Promartynovia, based upon a single wing insignis Till. and protohymenoides Till. The latter is fragment (venicosta Till.). Better and more nearly the genotype of Martynoviella, which I have synony- complete specimens of the latter species in the Har- mized here with 1Martynovia. In regarding these vard collection convince me that Promartynovia is two species as representatives of different genera, really a Protorthopteron. Tillyard himself suggested Tillyard was unaware that one of them (insignis) that the precostal area in Promartynovia was prob- was based on a hind wing, and the other on a fore ably a remnant of a forniation carried over from a wing. Although cognizant of the similarity of their Protoblattoid ancestor. The genus Promartynovia venation, Tillyard was led to place them in separate will accordingly be considered in the following part genera by the difference in wing shape. Actually, of this series of papers. of course, this difference is due to the fact that the For reasons which will be given below, I consider fossils do not represent homologous wings. the genus Martynoviella to be synonymous with Martynouia, which thus becomes the only known Martynovia insignis Tillyard genus in the family. Figure 2 Genus Martynovia Tillyard Martynovia insigais Tillyard, 1932, Amer. Journ. Martynovia Tillyard, 1932, Amer. Journ. Sci. 23: 13. Sci. 23: 15. Martynoviella Tillyard, 1932, ibid., 23: 17. firewing: length, 12.5-20 mm.; width, 3-4.7 mm. Fore wing: long and moderately slender; costal Costal margin arched basally, slightly concave just space of moderate width basally, but narrowed beyond the origin of Rs; costal space broad basally,

FIGURE2. Ma~tynouiainsignis Till. A, fore wing; drawing based chiefly on specimen no. 4601, Mus. Comp. Zool. B, hind wing; drawing based mainly on specimen no. 4604, Mus. Comp. Zool., and the holotype. Lettering as in figure I. THE LOWER PERMIAN INSECTS OF KANSAS. PART 9 59

but very narrow before end of Sc; several prominent cal, the only differences between them being in wing costal veinlets; Rs with from 4-5 branches; usually shape. Two possible interpretations of these wings two cross-veins between each of these and MA and can be made; they might represent two species of YIP; CuA and CUP separating at about the same Martynovia, one slender and the other broad; or level as the point of divergence of Rs from M; 1A they might be fore and hind wings of the same and 2A well developed; 3A sometimes present near species. The latter interpretation is the one which base of wing. Cross-veins strongly developed. seems certainly correct, in view of the wing struc- Hind wing: length, 12.5-20 mm.; width, 4.2-6 mm. ture of Elmoa, the fore and hind wings of which Costal area broader than in fore wing and not so show the same differences as those existing between markedly concave; venation as in the fore wing. the wings in figure 2. Furthermore, it is also appar- The hind wing appears to lack a pterostigma; its ent from the condition in Elmoa and other Mega- absence in the holotype (hind wing) can hardly be secoptera that the broader of the two wings is the due to its failure to be preserved, since the wing is hind one. in excellent condition. Holotype: no. 5423, Peabody Museum, Yale Martynovia protohymenoides (Tillyard) University. This is a hind wing, not a fore wing, as stated by Tillyard. Figure 3 In the Harvard collection there are three speci- Martynoviella protohymenoides Tillyard, 1932, Amer. mens of this insect, all from the upper layer of the Journ. Sci. 23: 18. limestone and collected at Elmo by F. M. Carpenter: Fore wing: length, 12.5-17 mm.; width, 3-3.5 no. 4600ab, and no. 4601ab, complete and well mm.; much more slender than that of insignis; preserved fore wings; and no. 4604ab, a hind wing, costal margin only slightly arched basally and almost about two-thirds complete, lacking only a part of straight beyond this part; costal space not narrowed the costal area and most of the actual posterior until about mid-wing; pterostigma as in insignis margin. though not extending so far proximally; Sc more Tillyard's drawing of the wing of insignis (1932, remote from R1 than in insignis; costal veinlets and fig. 5) is incorrect in two important respects: in the other cross-veins apparently fewer in number than first place, the L'obsolescent costal vein" which he in insignis; Rs with 34branches; CUP separating represents in solid lines in one illustration (fig. 5) and from R + M much further proximally than the in dotted lines in another (fig. 6), does not exist in point of separation of CuA; 3A is not preserved in the type nor is it present in the Harvard specimens. any specimen; cross-veins very weak. Hind wing Tillyard himself was apparently doubtful of its unknown. presence, for he states (p. 16) that he was unable to Holotype: No. 5424, Peabody Museum, Yale satisfy himself of the course of this vein. Such a vein University. This consists of a nearly complete fore does not occur in the Megasecoptera, though it is wing; only about the proximal fifth of the wing is found in many Neuroptera. In the second place, missing. the "obsolescent Cu2," represented in his figure by In the Harvard collection there are three speci- dotted lines, is also not present in any of the speci- mens of protohymenoides, all from the lower layer of mens, including the type. Of this vein he states the limestone and collected by F. M. Carpenter: that he could not be quite certain even of its exist- no. 4597ab, the distal half of a fore wing; no. ence. I am at a loss to explain why Tillyard did not 4599ab, a complete and well preserved fore wing; recognize as CUP (Cu2) the vein which he has and no. 4558ab, a nearly complete fore wing. designated Culb; this is concave, as CUPshould be, This insect is obviously very similar to insignis and it occupies the normal position of this vein. and differs from it in such minor characteristics as Since no complete specimens of insignis or proto- the shape of the wing and of the costal space. In hymenoides are known, I have based my conclusion separating protohymenoides from insignis generically, on the identity of fore and hind wings of these two Tillyard was misled by the fact that he had only the species on comparisons with the wings of Elmoa. fore wing of one species and the hind wing of the It is at once clear from an examination of the two other. His figure of the type of protohymenoides wings shown in figure 2 that their venation is identi- (1932, fig. 7) is not entirely correct. He has drawn hlart,ynov (1938b, p. 39), without having seen any more of the proximal part of the wing than is actually specimens of Martynoviidae, suggested on the basis of preserved and his restoration of the base of the the published figures that Tillyard's Culb (CuA2) was wing does not agree with the structure shown in the really CUP. IIarvard specimens. The pterostigma is faintly but 60 CARPENTER clearly indicated in the holotype, though it is not conclude, I believe, that the Martynoviidae do not mentioned by Tillyard. It is more distinctly pre- belong with the Neuroptera and have no relation- served in one of the Harvard specimens (no. 4598). ship with them. This view has already been ex- pressed by Martynov (1938b), though he did not present reasons for his co~iclusion. As previously mentioned, the family Martynovi- We are next faced with the problem of determin- idae was placed by Tillyard in the suborder Sialoidea ing the true affinities of the Martynoviidae. Of of the order Neuroptera. TJnfortunately, Tillyard the Permian insects now known the Elmoidae seem gives no reasons for his decision that Martynovia to be the closest to them, in wing structure, at least. and ,Vartgnoviella were Neuropterous, apparently All of the main veins have similar origins in both assuming that their systematic position was obvious. groups, and the short subcosta occurs in both. The However, in his discussion of the phylogenetic chief differences between them are in the coalescence relationship between the Martynoviidae and the of MA and Rs and the presence of a pterostigma in existing Sialoidea, Tillyard concluded that the the Martynoviidae; in the Elmoidae MA is free from Martynoviidae were in no sense ancestral to any Rs and a true pterostigma is apparently absent. Recent members of the order. Apparently, there- However, these two characteristics are the ones fore, he recognized the existence of great differ- which distinguish the Elmoidae from the other known ences between the living Sialoidea and the Marty- families of Megasecoptera. Consequently, the

FIGURE3. MartynozYia protohymenoides Till. Drawing of fore wing, based chiefly on specimen no. 4599, Mus. Comp. Zool. Lettering as in figure 1. noviidae. He explained these differences by the Martynoviidae are intermediate in wing structure assumption that the Martynoviidae were an end between the Elmoidae and certain other Mega- group of Neuroptera which arose from Carboniferous secoptera. It is my conviction, therefore, that the stock. In my opinion, however, the differences are Martynoviidae are Megasecopterous, belonging to simply due to the fact that the Martynoviidae are the suborder Eumegasecoptera; and that their not at all related to the Neuroptera or to any other systematic position is not far from that held by the holometabolous insects. Elmoidae. A comparison of the wings of the Martynoviidae The pterostigma of the Martynoviidae is of much with those of the Megaloptera shows that whatever interest in this connection. As already noted, this resemblances exist (such as the branching of Rs) pterostigma, although only slightly thickened, is are entirely superficial; whereas the numerous dif- very suggestive of that in the suborder Protohy- ferences are of a more fundamental type. The menoptera (Megasecoptera). In the latter insects, short subcosta of the Martynoviidae, for example, the area between R1 and the costal margin is is in direct contrast to the long one characteristic of thickened and pigmented for a considerable dis- the Megaloptera (Sialoidea), in which the distal tance. I do not believe, however, that the similarity part of the subcosta is always associated with the of the pterostigma in these two groups is sufficient radius in the pterostigrnal region. Another essential to justify the reference of the Martynoviidae to the difference is the presence of the extensive and Protohymenoptera, for they have much more in thickened pterostigma in the Martynoviidae; no common with the Eumegasecoptera. Nevertheless, such structure is present in the Sialoidea. Without they were probably not far from the line which gave enumerating further differences of this sort, we may rise to the Protohymenoptera. Incidentally, I wish THE LOWER PERMIAN INSECTS OF KANSAS. PART 9 61 to point out that although no true pterostigma has These eleven specimens are of much significance, been seen in the wings of the Carboniferous Mega- however, since they constitute the earliest geological secoptera, one might well have been present; all or record of the order. nearly all of the Carboniferous species are preserved It is a fact well known to entomolonists- that be- in black or very dark gray shale, in which such cause of their great diversity of structure the Neu- detailed structures as a pterostigma are obscured. roptera defy a concise definition. Even the wing Tillyard was obviously aware of the similarity venation in the order is of such a bewildering variety between the Martynoviidae and the Protohymenop- that taxonomists have found it impossible to draw tera, for he says of Martynoviella (1932, p. 23): up venational limits of the group. This diversity "There is here, surely, something more than a of venation leads to a serious problem in the case of purely chance resemblance to Protohymenoptera. the fossil Neuroptera, of which only the wings are While it can scarcely be doubted that this genus is usually preserved. At least some uncertainty neces- Sialoid, the strongly formed veins and cross-veins, sarily attends the determination of isolated wings with the obsolescent chaetotaxy, the general scheme as Neuroptera, especially specimens from Palaeozoic of venation and the curious arrangement of the strata. I emphasize this here because I am convinced cross-veins, strongly suggest the possibility of an that several of the insects from the Elmo limestone affinity between the two orders. And if, as may be which Tillyard has described as Neuroptera are not possible (though I cannot accept Carpenter's sup- even related to the order; and consequently that posed proof of it, of which more later) the Proto- the conclusions which he has based on these speci- are descended from Megasecoptera, mens are not valid. then we may also have to reconsider HandlirschJs Four of the Neuroptera which Tillyard described view that the Neuroptera are similarly derived." *from the Yale collection were placed by him in the The latter part of this statement is of much interest, suborder Planipennia (Neuroptera, s. s.) and four since four years later (1936) Tillyard accepted in others in the suborder Sialoidea (Megaloptera). full the thesis that the Protohymenoptera were The following list includes all of these insects: Megasecopterous (1936, p. 450). I do not for a Suborder PLANIPENNIA moment believe, however, that we are thus obliged Fam. PERMOBEROTHIDAE to give serious consideration to the derivation of the Permoberotha villosa Till. (1932) Neuroptera from such highly specialized insects as Permoberotha convergens Till. (1937) the Megasecoptera. As Tillyard himself pointed Dictyobiella nervosa Till. (1937) out (1932, p. 29), if there is one thing that appears Permobiella perspicua Till. (1937) certain it is that the two orders and Suborder MEGALOPTERA(Sialoidea) Neuroptera have come from a common stem, and Fam. Choristosialidae that this type must have combined in itself the Choristosialis enigmatica Till. (1932) more archaic characteristics of both the Meropoidea Fam. Martynoviidae (Mecoptera) and Sialoidea. Such an ancestor is ~Vartynoviainsignis Till. (1932) almost certain to be found among the Orthop- ~T~artynoviella protohytnenoides Till. teroids (probably Protorthoptera), not among the (1932) Megasecoptera. Prornartynovia venicosta Till. (1937) Martynov (1938b), who has expressed doubt of A study of both Yale and Harvard specimens of the Neuropterous affinities of the RIartynoviidae, these species has convinced me that none of those has suggested that the family might he allied to the described as Sialoids are really Neuropterous. The Archodonata, an order estal-~lishedby 11im for certain genera Martynouia and AIInrtyrioviella belong, in my Permian insects related to the . The addi- opinion, to tile order Megasecoptera; the reasons for tional fossils described in the present paper, however, this conviction have already been given above, in appear to eliminate this view, which was in fact the account of additional RIegasecoptera. The genus proposed by him merely as a possibility, pending the Pron~art~novia,which was based upon a srnall frag- study of further specimens of Martynoviidae. ment of a wing, is clearly Ortliopteroid; the evidence for this conclusion has been obtained from several Order NEIJROPTERA complete wings in the Harvard collection and will be The order Neuroptera is very sparsely represented discussed in the subsequent paper in this series. in the Elmo limestone. There are but two specimens Choristosialis should, I believe, be regarded as unquestionably referable to this order in the Yale hlecopterous until evidence to the contrary has been collection and only nine in the Ilarvard collection. found. The unique specimen on which the genus mas established consists of less than half of one wing. the affinities of Permobiella, for, as Martynov has Tillyard hi~nselfexpressed some doubt of tlie Neu- observed (1938a, 1938b), it belongs without doubt ropterous affinities of this insect, pointing out that to the Caloneuridae, an Orthopteroid family which it had certain Mecopterous features, such as the Tillyard apparently overlooked completely when he dichotomous branching of Rs and M. The latter described the fossil. Hence, Tillyard's rejection of seems to me to have far rnore significance- than the hlartynov's concept of the Neuropterous venation presence of possible Sialoid characteristics, such as was based on evidence which really did not exist. the costal veinlets, which incidentally are also pres- Since I have found, as indicated elsewhere (1940), ent in the generalized Mecoptera (suborder Pro- that Martynov's interpretation explains very satis- tomecoptera). If Choristosialis were regarded as a factorily the venation of the existing Neuroptera, Sialoid, it would be the only representative of the I have used it here. group in the Lower Permian; hence, in view of the incompleteness of the type specimen and the unques- Suborder PLANIPENNIA(Neuroptera, s. s.) tionable existence of true Mecoptera in the Lower Family PERMOBEROTHIDAE Permian, I prefer to consider it a Mecopteron. . Of the insects which Tillvard described as Plani- Wings subequal. Fore wing: costal area broader pennia, Permobiella is beyond doubt, as already than in hind wing; Sc terminating on costal margin, shown by Martynov (1938a, 1938b), a relative of the a series of costalveinlets connecting it to the anterior Orthopteroid family Caloneuridae, recently placed margin; pterostigma absent; Rs arising from R near by him in a new order, Caloneurodea. My account the base of the wing; MA coalesced with Rs for a of this genus, with descriptions of related genera, short distance proximally, then continuing as a will be found below under that order. The family nearly straight vein almost to the very apex of the Permoberothidae, however, almost certainly belongs wing; MP coalesced proximally with CuA for a short to the Planipennia. Only one valid species of this distance, then continuing close to MA and parallel family, Pemoberotha villosa Till., has so far been to it; between MA and MP there is a distinct groove, found in the Elmo limestone, and if its affinities have as in Recent and other families of been correctlv determined, it is the oldest record of Neuroptera; Cu forking at base into CuA and CUP; the order Neuroptera. CuA coalesced with MP for a short distance; CuA2 The venational terminology used in the following arising before the separation of MP from CuA; account of the Permoberothidae requires an expla- CUP is a faint but distinct, straight vein. Hind nation. In 1928 and again in 1932 Martynov ex- wing: Sc close to R and terminating on R1 not far pressed his conviction that the vein previously from mid-wing; Rs arising from R as in fore wing identified as the media in most existing Neuroptera and coalesced with MA for a greater interval than was in reality only a posterior branch of the media in the fore wing; Rs forked as in the fore wing; MA, (i. e., MP), and that the anterior branch had been MP, CuA and CUP essentially as in the fore wing; identified as part of the radial sector. Tillyard in groove present between MA and MP. his first paper on the Neuroptera of the Elrno lime- Little is known of the body structure: the abdo- stone (1922) accepted this interpretation and at- men was slender and terminated in a pair of short tempted to apply it to Permoberotha. The lettering cerci. of the veins in his drawing of the wing follows this The family Permoberothidae is so far represented terminology, but through some oversight his descrip- by only one genus. tion of the venation in the text gives a different interpretation, more nearly following that of Corn- Genus Per~~~oberothaTillyard. stock (1918). In his second paper (1937) on the Permoberotha Tillyard, 1932, Amer. Journ. Sci., 23: Neuroptera of the Elmo limestone, Tillyard rejected 24; 1937, ibid., 33: 108 Martynov's interpretation, which he had previously Dictyobiella Tillyard, 1937, ibid., 33: 104 accepted, and returned to Comstock's view, so far Fore wing: slender; costal area narrow basally, as the radial sector and the media were concerned. but broadened above the origin of Rs; recurrent But he did this solely on the evidence (1937, p. 102) humeral vein absent; costal veinlets not forked; Rs which he thought was provided by another Neurop- forked at about mid-wing or beyond, its branches teron from the Elmo deposit, Permobiella perspieua somewhat zigzagged; CuAl forked at least once; 2A Till.; this incomplete wing he believed to be a terminating on 3A. Cross-veins are numerous and I'lanipennian "reduced to its lowest terms." Unfor- not arranged in gradate series. Hind wing: costal tunately, Ile was far wrong in his determination of space with well developed veinlets; Rs forked as in THE LOWER PERAIIAN INSECTS OF KANSAS. PART 9 63 the fore wing; 26 terminating on wing .margin; veins. The fore wing liad well developed rnacro- cross-veins as numerous as in the fore wing, not in trichia on the main veins, their large sockets being gradate series. visible in all specimens. Microtrichia are discernible Genotype: Permoberotha villosa Tillyard. This is on the fore wing of the holotype. the only species of the genus known at present. Hind wing: length, 8 mm.; width, 2.8 mm. More oval than fore wing and much broader distally than Pennoberotha villosa Till. basally; Sc closer to Rl than to the anterior margin, with about 16 well developed costal veinlets between Figure 4 Sc and the margin, and between Rl and the margin. Pernzoberotha villosa Tillyard, 1932, Amer. Journ. The branching of Rs and CuAl shows the samevaria- Sci., 23 : 25 tion as in the fore wing. Permoberotha conuergens Tillyard, 1937, ibid., 33 : 109 The body structure of this insect is very incom- Dictyobiella nervosa Tillyard, 1937, 33: 105 pletely known and is preserved only in the holotype Fore wing: length, 9 mm.; width, 2.8 mm. Ante- of Dictyobiella lzervosa (no. 15592, in Yale collection). rior margin slightly concave; 10-12 costal veinlets, The length of the entire body is 6.5 mm., a little less and 7-9 between Rl and the costal margin; a single than that of the fore wing. The abdomen is slender, well developed cross-vein between Sc and R just 4 mm. long and 1mm. wide. Very little can be made before the origin of Rs; forking of Rs somewhat out of the head and thorax; Tillyard has described variable; in the holotype R2 + 3 and R4 + 5 the shape of the head, eyes and clypeus from this separate in the distal third of the wing, but in one fossil, but I have been unable to identify these of the Harvard specimens (No. 4605) they separate structures. The abdomen terminates in a pair of before mid-wing; in none do either of these branches cerci. These were not entirely exposed when Till- of Rs apparently divide again; CuAl shorn little yard described the insects and are therefore incor- variation in forking; in all clearly preserved speci- rectly depicted in his figure (1937, fig. 9a). They mens each main branch of CuAl forks again at the are much longer than he thought and consist of 4 or margin. In the holotype, an incomplete wing, there 5 segments. There is no terminal bristle, as drawn are about 56 cross-veins; and in the best of the by him; the structure which he interpreted as .a Harvard specimens (No. 4605) there are 64 cross- bristle was an exposed part of the rest of the cercus.

FIGIJRE4. I'cr~,~ob~rot/itr~~zllo,str Till. A, fore wing; drawing based mainly on specimens no. 4605, Mus. Comp. Zool., ant1 tlie l~olotype. 13, hind wing; cl~anirlgbasccl n1:linly on spc~cin~cnno. 4575, RIus. Comp. Zool. R2 + 3 :inti It11 + 6, hranches of Its; other let tc,ring as in figure 1. Holotype: Ko. 5-126 and its counterpart (155'35) betwren- the wings of tlie llerotllidae arid Permo- Peabody Museum, Yale University. are so great I find it difficult to make In the Harvard collection there are nine specimens comparisons. The following summary lists the of this insect, nos. 4573-4580 and 4605. Six of these points of difference between these wings which seem are from the upper layer; four (4572, 4576, 4578, to me to eliminate any possibility of the relationship 4579) are hind wings and the rest are fore wings. suggested by Tillyard. All were collected by F. M. Carpenter in the Harvard 1. Permoberotha lacks any sign of a pterostigma. quarries at Elmo, Kansas, in 1927, 1932 and 1935. Since a pterostigina of some kind is present in all Permoberotha convergens Till. was based upon the the Hemerobioidea, including the Berothidae, the counterpart (obverse) of the holotype of villosa. absence of it can only be regarded as a specializa- This is not readily apparent from Tillyard's figures tion-unless of course we consider the very improb- of these two insects (Tillyard, 1932, fig. 11; 1937, able supposition that Permoberotha is ancestral to fig. lo), but when I examined the specimens them- all Hemerobioidea. selves at the Peabody Museum, I was struck by the 2. The subcosta of Permoberotha terminates at exact correspondence of cross-veins; and on placing about mid-wing. In the Berothidae and all other the two fossils together I found that they fitted existing Neuroptera (Planipennia), as far as I am perfectly. Tillyard's figure of the obverse specimen aware, the subcosta terminates nearly at the apex (cor~vergens)is very inaccurate and highly diagram- of the wing. matic. The numerous cross-veins which he shows 3. The radial sector of Permoberotha consists of between CuA, CUPand the anal veins simply do not only two branches, in contrast to the multibranched exist in the fossil. radial sector of the Berothids and nearly all existing Dictyobiella nervosa is likewise a synonym of vil- Planipennia. losa. Tillyard's figure of its wing does not indicate 4. The branches of Rs and Cu are strongly zig- this, but, as he mentions, his drawing was a com- zagged, a specialization which is not found in any posite one, based upon both fore and hind wings. Berothid. The hind wing venation is represented by the short 5. MA is unbranched and close to MP in Permo- Sc and the fore wing venation by the structure of berotha; the proximity of MA and MP is a special- Rs. The cubitus is incorrect for either wing and the ization not known in existing Neuroptera. zigzagged R4 + 5 (Tillyard's notation) shown in 6. MP in Permoberotha is coalesced basally with his figure and used for generic diagnosis, does not CuA in both wings, another specialization which is exist in the specimen. In the type specimen of not known to me in existing Planipennia. nervosa the two fore wings and one hind wing are 7. Sc in the hind wing of Permoberotha is very superimposed so that most of the veins cannot be short, terminating on R1 well before the middle of distinguished, though enough is clear to enable the wing; such a structure of Sc is unknown in Recent recognition of the venation of villosa. The second Neuroptera. hind wing is preserved out-stretched to one side and These differences have led me to the opinion that its venation is distinct except at the base. It is the family Permoberothidae is much too highly identical with the hind wing shown here in figure 4, specialized to be ancestral to the Berothidae, or for except for the position of cross-veins and other indi- that matter to any other existing family of Neurop- vidual features. Unfortunately Tillyard assumed tera. It is more likely a specialized offshoot of a that the short Sc and other hind wing structures group of Upper Carboniferous Neuroptera, having were characteristic of both wings and made them little phylogenetic relationship with existing mem- generic. It is noteworthy that Permoberotha villosa bers of the order. This conclusion is substantiated is the oilly Permian Neuropteron so far described of by a comparison of Permoberotha with the Upper which both fore and hind wings are known. Permian Neuroptera which have been described from Russia arid Australia (Martynov, 1932; Till- yard, 1926). All of these Upper Permian species, of Since Permoberotha is geologically the oldest which seventeen have been described. are far more Keuropteron known, its phylogenetic position is of generalized than Permoberotha and show numerous much interest. Tillyard was of the opinion that characteristics found in Recent families, such as the there was "no valid reason why the genus Permo- Berothidae, , and . The berothn should not be regarded as representative of Permoberothidae are certainly far removed from the ancestral stem of the Berothidae" (1932, p. 28). these other Permian types. I do not agree with this view. In fact, the difference The real significance of Permoberotha is not to be THE LOWER PERMIAN INSECTS OF KANSAS. PART 9 65 found in the evidence it provides for the relationship and connected to it by several long cross-veins; a or evolution of existing Neuroptera; but (provided well developed cross-vein joins M (or MA) to R1 it is a Neuropteron) in the evidence it furnishes that (or Rs) before the coalescence of MA and Rs, the order had been established long enough by the forming a conspicuous triangular or quadrilateral time of the lower Permian to include such highly cell. Hind wing: similar to fore wing, except for the specialized members. differences already noted. The body structure is unknown. Order RAPHIDIODEA Genotype: Permoraphidia americana Tillyard. A single species of Raphidiodea, Permoraphidia americana, was described by Tillyard from the Elmo Permoraphidia americana Tillyard limestone in 1932. This insect is obviously a very rare one in the formation; the holotype is the only Figure 5 specimen of it in the Yale collection and there are Permoraphidia americana Tillyard, 1932, Amer. but six additional specimens in the Museum of Journ. Sci., 23: 8. Comparative Zoology. An undescribed species of Fore wing: length, 9.5-10.5 mm.; width, 2.3-2.6 the same genus is also present in the Harvard col- mm.; elongate and oval; costal margin only slightly lection. These new fossils fully substantiate Till- arched, apex somewhat pointed; recurrent humeral yard's conclusion on the affinities of the Permora- vein extending only to about the level of separation phidiidae and add materially to our knowledge of of CuA and CUP, with about 6 veinlets between it their structure. and the anterior margin; 4-8 costal veinlets; pterostigma rather broad, of uniform density; Family PERM~ several subcostal veinlets present, usually four in Wings subequal; the membrane lacks microtrichia pterostigma; Rs originating at about mid-wing; the and apparently macrotrichia, but it is not glassy, as amount of coalescence between Rs and MA is about in the existing Raphidiodea. Fore-wing: a recurrent equal to the length of the free part of Rs before the humeral vein, resembling that of some Neuroptera, coalescence; Rs forked to level of middle of ptero- is present; this vein may be a submarginal costa, as stigma; both terminal branches of Rs forked again it was interpreted by Tillyard. Costal veinlets well at margin; MA with two or three terminal branches; developed and about as numerous as in Raphidiidae; NIP with one or two well developed terminal pterostigma 'well developed, traversed by several branches; amount of fusion between M and CuAl cross-veins; R1 unbranched; Rs arising at about variable, but it is at least equal to the interval of mid-wing, with at least two branches; M coalesced coalescence of Rs and MA; CuAl forked to about with R basally and with CuAl for a short distance, half its length; CuA2 simple, unbranched; 1A close the free piece of CuAl before this coalescence being and parallel to CUP;2A close to 1A proximally, but oblique; MA and MP diverging distally to the diverging from it distally; 3A short and close to the separation of CuAl from M; MA coalesced with Rs posterior margin; between all anal veins is a variable for a short distance; Cu free from M basally, shortly number of cross-veins, mostly curved or sigmoidal. dividing into CuA and CUP; CuA forking before Hind wing: all of the hind wing is known except the mid-wing; CuAl diverging anteriorly and coalescing distal part of CUP and the anal veins. The largest with 91 for a short distance; CUPweak, unbranched; fragment is 8 mm. long and 2.5 mm. wide; in all la, 2A, 3A unbranched; cross-veins more numerous probability the wing was a little shorter than the than in Raphidiidae, especially in anal area. Hirid fore wing, though about as broad. Costal area wing: slightly smaller than fore wing and with nar- narrower than in fore wing; Sc terminating as in rower costal area; recurrent humeral vein absent; fore wing; pterostigma slender; Rs arising further venation similar to that of fore wing (anal area proximally than in the fore wing, M forked and unknown). coalesced with CuAl as in fore wing. The family is represented by the single genus Holotype: No. 1522ab, Peabody Museum, Yale Permomp hidia. University. In the Harvard collection there are six specimens Genus Permoraphidia Tillyard of this insect from the Elmo limestone, three from Permoraphidia Tillyard, 1932, Amer. Journ. Sci. 23: each layer: no. 4581, a complete fore wing, without 6. distortion and very well preserved, except for the Fore wing: Rs with at least two forks, each having cross-veins; no. 4582ab, distal third of a fore wing, terminal branches; Rs widely separated from Rl well preserved; no. 4584ab and no. 4585ab, fore CARPENTER

FIGURE5. Permoraphidia americana Till. A, fore wing; drawing based mainly on specimens no. 4581 and 4584, Mus. Comp. Zool. B, hind wing; drawing based on specimens no. 4586 and 4583, Mus. Comp. 2001. M, common stem of MA and MP; other lettering as in figure 1. wings, incomplete but well preserved; no. 4586ab, a of Permoraphidia. Macrotrichia, which are well hind wing, fairly well preserved; 4583ab, distal developed in Recent Raphidiodea, are not preserved third of a hind wing. In the Sellards collection there in any of the specimens of americana. is one nearly complete fore wing (no. 113). The drawing of the fore wing in figure 5 is based Permornphidia gmndis, n. sp. mainly on specimen no. 4581, with the addition of some cross-veins shown better in specimen no. 4584. Figure 6 It should be noted that Tillyard's figure of the fore Fore wing: length of the preserved part of wing, wing of arnericnna (1932, p. 6) is incorrect in several 15 mm.; estimated length of entire wing, 17 mm.; details, such as the structure of the "recurrent width of preserved part of wing, 3 mm.; estimated humeral veinJ' (termed the costa by Tillyard) and width of entire wing, 4 mm. Costal space much the proximal part of the posterior wing margin. I broader basally than in nmericana; recurrent hu- have identified two of the fossils (nos. 4586, 4583) meral vein also longer than in americana, with about as hind wings of this species solely on the basis of ten veinlets leading from it to the costal margin; their venation, no complete specimens of americana pterostigma slender, divided into two parts, the having been found. The wings mentioned have a anterior part being thicker than the posterior part; venation identical with that of the type of americana, Rs forked as in americana; MA coalesced with Rs except for differences which in general occur between with a short distance; MP and CuA as in americana; the fore and hind wings of existing Raphidiodea. CUPand anal veins unknown. The wings of americana were apparently very Holotype: no. 4589, Museurn of Comparative delicate. This is indicated by the fact that all but Zoology; collected in Elmo limestone (upper layer) one of the eight wings of this insect which have been at Elmo, Kansas (I?. M. Carpenter). collected are folded or torn. It is most interesting This insect was obviously a close relative of to find this true of Permoraphidia, since the wings of americana, and I have assigned it to the same genus, the existing Raphidiidae are also fragile. Appar- even though the structure of the hind wings is ently, however, the Permoraphidiid wings were not unknown. It had almost twice the wing expanse of glassy, as in the Recent members of the order. Such americana, and the fore wing was probably much a glassy texture is clearly preserved in the fossils of broader than that of the genotype. The most the Elmo limestone, as in the Megasecoptera and peculiar feature of grandis is the pterostigma, the Odonata; but is not even suggested in the specimens anterior part of which is thick and swollen, whereas THE LOWER PERMIAN INSECTS OF KANSAS. PART 9

FIGURE6. Permoraphidia grandis, n. sp. Fore wing; drawing based on holotype. Lettering as in figure 5. the rest, though pigmented, is membraneous, re- CuAl with M in the hind wing of Permoraphidia is sembling the entire pterostigma of americana. of much interest, for the possibility of such a coa- lescence in the hind wings of existing Raphidiodea has not previously been suggested. I am now Unlike the Mesozoic and Tertiary genera of convinced that this coalescence (CuAl and M) does Raphidi~dea,~Permoraphidia is decidedly general- take place in the Raphidiidae and Inocelliidae. An ized in comparison with the Recent members of the examination of the hind wing of any of the existing group, and it has a significant bearing on the prob- genera of snake- shows that the vein which I lem of the interpretation of their venational hom- have previously regarded (Carpenter, 1936, fig. 1) ologies. Martynov suggested over ten years ago as the posterior branch of MP is strongly convex, as (1925) that the anterior media, as well as the pos- CuAl should be. A comparison of such a wing terior media, was present in the existing Raphidi- (Fig. 7) with the hind wing of Permoraphidia indi- odea. The vein which had previously been deter- cates that the obscurity of the coalescence in the mined as the most proximal branch of Rs he believed former has resulted from the distal migration of the to be the anterior media, which was coalesced proxi- point of separation of CuAl from M. This migra- mally with the stem of Rs. The evidence for this tion has progressed so far that CuAl in existing was derived mainly from analogies with the wings of Raphidiodea is now fused proximally with the base Neuroptera (Planipennia) . The venation of Pemno- of MP. The snake-flies (Mesoraphidiidae), raphidia, as Tillyard has shown, fully substantiates it must be admitted, are fully as specialized as the Martynov's conclusions. In the fore wing of Recent species in this respect, and do not show an Perrnoraphidia the actual origin of MA and its intermediate condition, as might be expected. coalescence with Rs are very clear; so, too, are the Although Permoraphidia thus aids in the determi- origin and coalescence of CuAl with M. In the nation of the venational homologies in Recent hind wing of Permoraphidia, which was not known Raphidiodea, I do not believe that this genus, or to Tillyard, the structure of MA and CuAl is prac- the family which it represents, is ancestral to the tically identical with that of the corresponding veins Mesoraphidiidae, Raphidiidae or Inocellidae. The in the fore wing, except that they originate somewhat chief difficulty in the way of such a relationship is nearer the base of the wing, as in the hind wings of the presence of the strong, recurrent vein in the existing Raphidiodea. The distinct coalescence of humeral angle of the fore wing. This vein was identified by Tillyard as "the last vestiges of a truly See Martynov, 1925; Carpenter, 1936. archaic costal vein homologous with that in Permian

FIGURE7. Hind wing of Agulla assirnilis (Alb.) (Recent). Lettering as in kure 5. Plectoptera." Interpreted in this way, it would be Neuroptera. The present paper is therefore the a generalized structure, lost in the subsequent evolu- first account of the Caloneurodea as such in the tion of the order. In my opinion, however, it is Elmo beds, and it is based upon the specin~eriscon- more likely the homologue of the "recurrent humeral tained in both the Yale and Harvard collections. vein" which is found in many Neuroptera, and The characteristics of the order Caloneurodea which is formed, apparently, by the aligmnent of have been thoroughly discussed by Martynov in a several costal veinlets. The absence of the recurrent posthumous publication (1938b), but since his paper vein in the hind wing of Permoraphidia, as in the was printed in Russian, I include here a summary of Neuroptera, supports this view, rather than Till- these characteristics, including several details yard's; for the submarginal costa (costal brace) of furnished by the new fossils. the Permian Plectoptera is as well developed in the The body structure of the Caloneurodea is only hind wing as in the fore. Tlllyard's reference to the partially known.6 The head was of moderate size condition in Oliarces is rather puzzling. I have and apparently hypognathous; the antennae were examined the unique specimen by which this genus very long and slender; the first antenna1 segment was is known in the Museum of Comparative Zoology, large and swollen; the structure of the eyes is un- and find that the recurrent vein there is like that of known. The prothorax was about as long as wide. other Planipennia-apparently formed of aligned The abdomen was of moderate size, rather slender, veinlets. For these reasons, I consider the recurrent and terminated in a pair of short cerci. The legs humeral vein in the fore wing of Permoraphida to be were slender, often tenuous; the posterior pair were a specialized, rather than a generalized character- longer than the others but were not modified for istic. jumping. The tarsal segmentation is unknown. The fore wings of the Caloneurodea were at least Order CALONEURODEA moderately slender and were often very elongate. This order was established by Martynov in 1938 The hind wings were similar to the fore wings in (1938a) for a series of fossils formerly placed in the shape and venation; they were a little broader than Protorthoptera. The following families were in- the fore wings, but did not possess an enlarged anal cluded by him in the new order: area.' The fundamental pattern of the venation seems to have been much the same in all families, so far as known. At first glance this pattern seems Caloneuridae (France; Pennsylvania) very simple and generalized, with MA, MP, CuA, Protokollaridae (France) and CUPpresent. However, a more detailed study Stenaroceridae (France) of the veins, especially with reference to their con- ? Emphylopteridae (France) vexity or concavity, reveals a most perplexing con- dition, and, in my opinion, a highly specialized one. The identity of the subcosta, radius, and radial Euthygrammidae (Russia) sector is obvious. Se is long, extending at least to Permobiellidae (Kansas) mid-wing and often to the apex; Rl is unbranched Synomaloptilidae (Russia) and terminates about at the apex; Rs usually origi- nates somewhat proximally of the mid-wing and In my opinion, the Protokollaridae and Stenaro- gives rise to a variable number of branches, from ceridae, both of which were based on single genera, two to many. The homology of the next three do not belong to the Calone~rodea.~The family longitudinal veins is more difficult to determine. Emphylopteridae, which was doubtfully included The convexity or concavity of all veins, except the by Martynov in the Caloneurodea, I also exclude from this order. However, in this paper I am erect- "Iartynov has published a good restoration of one ing three additional families of Caloneurodea: of the species from the Permian of Russia, though not Paleuthygrammidae, Pleisiogrammidae, and Ano- all the details shown in his figure are actually known malogrammidae. These are established chiefly for (1938b, p. 126, fig. 63). species in the Elmo limestone. Although Tillyard Zeuner has stated (1938, p. 69) that the hind wings described one 'species of Caloneurodea in the Yale of the Caloneuridae possessed an anal fan. I do not Collection from this formation, he placed it in the know any basis for this assertion. None of the speci- mens which I have seen, including those from Com- The reasons for this conviction will be discussed in mentry, show this anal fan and so far as I am aware no my revisional stJudy of the insects of the Commentry other mention of it occurs in the literature on the Calo- shales, now in preparation. neurodea. THE LOWER PERMIAN INSECTS OF KANSAS. PART 9 69

anals, is very strong; this is true of Sc and Rs, which assume that these particular veins have changed are concave, and of R1, which is convex. Below Rs through evolutionary processes from a concave posi- is a forked vein (labeled 1 in figure 8) which arises tion to a convex one or vice versa. I do not believe directly from the vein below it; this vein (1) and such changes have ever been demonstrated in the both of its branches are strongly concave, like Rs. insects (except perhaps in the case of elytra, where The vein from which it arises (labeled 2 in figure 8) the thickening of the wing introduces another fac- is very strongly convex; paralleling this vein and tor); and until that has been done, I consider situated just below it is a concave vein (3 in figure 8). Martynov's interpretation untenable. The follow- In his 1906 treatise on fossil insects Handlirsch ing identification of the veins is used here: vein 1 is figured the wing of the Carboniferous Caloneura and MP ( -), MA having been lost; vein 2 is CuA ( +), identified vein 1 (figure 8) as the media, and veins being coalesced proximally with MA; vein 3 is CUP 2 and 3 as branches of Cu. This is the obvious (-). I must admit, however, that one peculiarity interpretation of the veins, made without regard to of the venation of the Caloneurodea leads me to their convexity or concavity. Lameere, in 1917, have a suspicion of doubt about even this interpre- identified vein 1 of Caloneura as the media, vein 2 tation. The vein here designated as CUP, although as CuA ( +) and vein 3 as CUP ( -). This interpre- close to CuA for most of its length, diverges away tation agrees with the convexity and concavity of from it proximally and coalesces with the base of

SC-

FIGURE8. Culoneuru duwsoni Brongn. (Carboniferousof Commentry, France). Original drawing, based on types in the Museum dlHistoire Naturelle, Paris. Sc, subcosta; Rl, radius; lls, radial sector; MA, anterior media; MP, posterior media; CuA, anterior cubitus; CUP,posterior cubitus; lA, 2A, etc., anal veins; +, convex veins; -, concave veins. For explanation of (I), (2), (3), see text. , the branches of Cu. Martynov (1928) in his descrip- the vein below (here identified as IA), which shows tion of Euthygramma, and Tillyard (1937) in his neither convexity nor concavity. This basal coa- account of Permobiella (which he placed in the lescence is apparently a characteristic of the order, Neuroptera) interpreted the venation of their for it occurs in all genera, Carboniferous as well as respective genera in like manner. In 1938, however, Permian, which have been found. The difficulty of Martynov, after a survey of the Order Caloneurodea, the present interpretation is that CuA and CUP do proposed a new interpretation, designating vein 1 not originate from a common stem; instead, CUP as MA, vein 2 as MP, vein 3 as CuA, and the con- arises from the base of 1A. Such a common origin cave vein below the last as As Martynov of 1A and CUP does occur, of course, among living has noted, according to this interpretation, MA is insects, but its presence in the Caloneurodea adds concave, MP is convex, CuA concave, and CUP another specialization to a wing which at first glance convex. These are exactly the reverse positions of seems to be generalized. the homologous veins in other insects. Such an The anal veins of the Caloneurodea vary from interpretation can therefore be valid only if we five in the Carboniferous species to two in some of the Permian members. One of the most striking U~lfortunatelyMartynov based this interpretation features of the wings of these insects is the convexity on the structure of the genus Euthygaw~ma,which differs radically from all other Caloneurodea in having of the numerous cross-veins. In all known species vein 2 remote from vein 3, and in having the latter very the cross-veins are very heavy and are often stronger close to the concave vein below it,. See Mart'ynov, than the longitudinal veins. This peculiarity and 1938a, figure la, p. 71. the proximity of CuA and CUP (or CUP and 1A in 70 CARPENTER

Euthygramma) are the two chief characteristics Ily 3 anal veins; 1A and 2A close together proximally, which the order may be recognized. diverging distally; 3A sigmoidally curved, termi- The classification of the species of Caloneurodea nating at about mid-wing; cross-veins numerous, into families is at present very uncertain. Since we usually straight, sometimes sigmoidal. The fore know but few species within the order, any classi- wing (and perhaps the hind also) has a peculiar, fication is likely to be more convenient than natural. granular texture. This is indicated in only one, Hence, whatever family classification is now adopted splendidly preserved wing (no. 15572ab), but the will undoubtedlv need to be modified as additional absence of the granular structure in other fossils is species become known. Inasmuch as the new almost certainly a matter of poorer preservation of genera from the Elmo limestone do not fit satis- the insects. The granulation is apparently caused factorily into families already established, I have by numerous minute dark spots, resembling the been compelled to erect three new families for them. bases of microtrichia. No macrotrichia or micro- A survey of all described genera of the order, both trichia are visible on any of the wings. Carboniferous and Permian, reveals two striking Hind wing: only a little shorter than the fore wing; facts: 1) all genera are very similar in general pattern the costal space is somewhat narrower than that of of the venation, and 2) all show intermediate condi- the fore wing, and hind margin is more curved tions of their generic characteristics. The pro- proximally. nounced upward arching of MA1 + 2 in the new The body structure is very little known. The genus Apsidoneura is illustrative of the latter. This head was of moderate size, the antennae and legs arching is present to a slight extent in several of the elongate, as in other Caloneurodea. Permian genera and especially in the Carboniferous This family is erected for two new genera and also Homaloneura. Another illustration is furnished by for several genera which Martynov placed in the the branches of Rs. These range from six or seven Euthygrammidae. The genus Euthygramma, on in the Caloneuridae to none in Euthygramma; but which the latter family was based (Martynov, 1928), the number of branches varies specifically to such is so different from the other genera included there an extent as to render it useless for generic diagnosis. that I believe it requires a separate family. These There are many other venational characteristics differences are chiefly the unbranched Rs and the which show the same tendency. So far I have found proximity of CUP and 1A, instead of CuA and it virtually impossible to erect the framework of a CUP, as in other genera. satisfactory classification of these peculiar insects. The family Paleuthygrammidae, which is known The affinities of the Caloneurodea are not very from the Permian of Russia and Kansas, includes clear. They are undoubtedly a part of the Orthop- the following genera : Paleuthygramma Mart ., teroid complex and may be, as Martqnov believed, T'ilviopsis Mart., Vilvia Zal., Pseudogramma, n. an offshoot of the early Protorthoptera. Zeuner has gen., and Apsidoneura, n. gen. Vilvia is readily suggested (1939) that the Caloneuridae might be distinguished by the density of the cross-veins. related to the ancestors of the Acridioidea, and also Paleuthygramma may be defined at present as has stated that the primitive acridioid family including those species with a forked Rs. Vilwiopsis 1,ocustopsidae could easily be derived from the has a radial sector with three branches, but I suspect Caloneuridae. I find it difficult to conceive of such that additional specimens of Paleuthygramma will a relationship on the basis of our present knowledge. show that generic separation because of the differ- Whether or not the Caloneurodea require permanent ence of one branch of Rs will not hold. The new separation from the Protorthoptera as a distinct genus Pseudogramma is here erected by me for order can only be decided after many more species aberrans Martynov, which was placed by its author and anatomical details are known than now. in Euthygramma. Martynov considered it probable that MP in this species was coalesced with Rs. In Paleuthygrammidae, new family view of the structure in additional Caloneuridae. however, I believe that such a condition is most Fore wing: long and slender; costal area of improbable and that MP is really unbranched, as moderate width; Sc long, terminating not far from in one of the Kansas species of Caloneurodea. This, the apex of wing; Rs arising near the middle of the of course, implies that the straight and nearly con. wing, with two or rarely three branches; MP tiguous veins in the wing are CuA and CUP. How- arising well before and forking beyond the origin of ever, in either case, i. e., whether MP is a simple Rs; CuA and CUPstraight and very close together, vein or is coalesced with Rs, a distinct genus appears terminating in the distal fourth or fifth of the wing; necessary for this species. THE LOWER PERMIAN INSECTS OF KANSAS. PART 9 7 1

Two genera of this family, Paleuthygramma and Paleuthygramma acuta, n. sp. ilpsidoneura, n. gen., occur in the Elmo limestone. Figure 9 Fore wing: length (holotype), 18 mm.; width, 3 Genus Palezithygrn?nrna Mart. mm.; costal margin slightly narrowed at very base; Pnleuthygramma Martynov, 1930, Ann. Soc. Pal. de Sc terminating before the level of the forking of Rs; Russie, 8: 42; 1938, Trav. de llInst. Palhnt., Rs with a conspicuous though shallow fork; some- 7: 126. times the alignment of cross-veins within the latter Fore wing: Sc terminating at about the level of fork gives the impression of a weak intercalated the fork of Rs, usually a little proximad to it; Rs vein between its branched; point of separation of arising at mid-wing, or slightly before it and forked MP from Cu variable, but always well proximal to dichotomously near the apex of the wing; MP deeply the origin of Rs; anterior branch of MP slightly forked, almost to half its length; 2A and 3A sinu- curved in its distal half; 1A close to 2A proximally; ously curved. The cross-veins are numerous, as 2A slightly curved sigmoidally; 3A arched near the characteristic of the family; most of them are base of the wing. Hind wing: length (holotype),

FIGURE9. Paleuthygramma ncuta, n. sp. A, fore wing; drawing based chiefly on holotype and specimen no. 4563, Mus. Comp. Zool. B, hind wing; drawing based chiefly on holotype and specimen no. 4554, Mus. Comp. Zool. Lettering as in figure 8. straight, but those in the distal part of the wing 17 mm.; width, 4 mm.; similar to the fore wing, tend to be sigmoidal. Hind wing: similar to the except for the differences noted under the genus. fore wing, but with a narrower and straighter costal Very little is known of the body structure of this margin; apex slightly more rounded than that of insect. The abdomen, preserved in three specimens fore wing; 3A more curved than in fore wing. (Peabody Museum, 15655; M. C. Z. 4587, 4547) is This genus was established by Martynov for the 9 mm. long and 2.5 mm. wide. One specimen single species tenuicornis Mart., from the Permian (M. C. Z. 4547) shows several terminal abdominal of Russia. The unique type consisted of a whole appendages. Two of these (3 mm. long) are seg- specimen, which showed the general body structure mented and are probably cerci; the others, included and most of the venation of one wing. A closely between the cerci, are incomplete and I have not related species is represented in the Yale and Har- been able to determine their precise form. The vard collections from the Elmo limestone. In fact, prothorax (hf. C. Z. 4587) is 2 mm. long and 2 mm. its venation is so close to that of tert7ricorni.s that I wide. The legs were obviously long and slender; have found it impossible to separate the two species the middle femur is 5 mm. long and the tibia is at generically, in spite of the great difference in their least 4 mm. long, since that much of a tibia is pre- geographic range. served without the distal end being evident. One $2 CARPENTER

hind tibia is preserved (M. C. Z. 4545); this is 9 which includes parts of all four wings. The differ- mm. long. The tarsal segmentation is unfortunately ences are indeed slight, as indicated above, and are not known. The body structure of acuta, as far mainly confined to wing shape. This difference in as known, was apparently much like that shown in shape between the fore and hind wings is almost Martynov's restoration of Paleuthygramma teaui- certainly a generic characteristic. cornis (1938, fig. 63), except that the legs of acuta The individual fossils show slight variation in were probably longer. venation. The place of origin of MP is subject to Holotype: hluseum of Comparative Zoology, No. slight variation as is also the depth of the forks of 4587ab, collected in the lower layer of the limestone Rs and MP; but the individual differences are not at Elmo, 1935 (F. M. Carpenter). This consists of comparable in extent to those in the Protoperlaria a partially complete insect; three wings extend along and most other groups of Permian insects (see the abdomen, and one fore wing, preserved only to Carpenter, 1935). about the middle, lies at right angles to the body. This specimen therefore shows part of a fore wing Apsidoneura, new genus and at least one complete hind wing, though the Wings subpetiolate and slender, though not so proximal part of the latter is obscured by the elongate as in Paleuthygramma; Sc terminating abdomen. almost at very apex of wing; Rs arising somewhat Paratypes as follows: Two in the Peabody nearer the base than the apex, with 2-3 branches; Museum, Yale University, collected in the Elmo MP originating at point of separation of Cu from R beds by C. 0. Dunbar: No. 15572ab, a fore wing, and M, and forking well beyond mid-wing; anterior very well preserved, but with the posterior part, branch of MP strongly arched towards Rs; CuA below lA, broken away; this specimen shows clearly and CUPstraight and very close together; CUPand the texture of the wing. No. 15571, a nearly com- 1A coalesced basally; 3 anal veins present; 1A plete fore wing, with apex missing. remote from CUP; 3A somewhat shorter than in Nine in the Museum of Comparative Zoology; all Pseudogramma and Palet~thygramma. collected at Elmo, 1927-1935 (F. M. Carpenter): This genus is known only from two isolated wings; No. 4563ab, basal third of fore wings, excellent both are probably fore wings. preservation (upper layer of limestone); No. 4554, a nearly complete hind wing, apex missing (lower Apsidoneura flexa, n. sp. layer); No. 4566, basal half of fore wing (upper Figure 10; plate 1, fig. 2 layer); No. 4558ab, basal third of fore wing, excel- Length of wing (holotype), 30 mm.; maximum lent preservation (upper layer); No. 4547ab, parts width, 6.5 mm. Costal margin straight, apex of wings and body, especially the abdomen (lower rounded, posterior margin convex; proximal third layer); No. 4560ab, distal third of wing, excellent of wing more slender than rest; costal space much preservation (lower layer); No. 4556, nearly com- broadened in proximal half, but narrowed distally; plete fore wing (lower layer); No. 4558ab, nearly Rlterminating at apex of wing; Rs curved anteriorly complete fore wing (upper layer). towards Rl opposite the upward curve in anterior In addition to these twelve type specimens, there branch of MP; MP straight from its origin to its are four other specimens in the Yale collection (Nos. fork; anterior branch of MP arching abruptly to- 15661, 1565513, 15657ab, 15656a) and twenty in the wards Rs for a short distance, them curving pos- Museum of Comparative Zoology. These make a teriorly; CuA and CUP are weak and twigged at total of thirty-five specimens in both collections. their terminations; 1A terminating about at level of Since none of these shows either a fore or hind fork of MP; 2A terminating just beyond level of wing entirely complete, the drawings in figure 9 origin of Rs; 3A well developed but close to wing have been based on several specimens. A com- margin. Cross-veins numerous, convex, the longer parison of figure 9 with that of Paleuthygramma ones sigmoidally curved. In the region of the tcnuicornis Mart., the genotype, reveals the close branches of Rs some cross-veins have aligned to similarity of the species. Probably a greater differ- form short intercalated veins. ence exists in the anal area, which is not preserved in Holotype: No. 4572ab, Museum of Comparative the genotype; but I do not believe that this will be Zoology, collected in the upper layer of the Elmo enough to require generic separation of the two limestone, 1932 (F. M. Carpenter). This is a species. splendidly preserved specimen. The correlation of the fore and hind wings of this Paratype: No. 4551, Museum of Comparative insect is made possible chiefly by the holotype, Zoology, collected in the lower layer of limestone at THE LOWER PEHh3IAN INSECTS OF KANSAS. PART 9

FIGURE10. Apsidot~euraJlexu, n. sp. Drawing of holotype. Lettering as in figure 8.

Elmo (I?. M. Carpenter). This is an incomplete is so different from that of other Caloneurodea that wing, lacking that part of the wing which is proximal I have been unable to allocate them to any described to the separation of MP from R + M. The wing family. They apparently represent a specialized is 23 mm. long as preserved and 6.5 mm. wide. branch of the order, characterized by almost com- The venation is identical with that of the holotype, plete loss of the third anal vein. I have tentatively except that Rs is 3-branched, R2 + 3 forking near placed both species in the single genus Pleisio- the apex. gramma, but it may eventually be necessary to This species unquestionably requires generic separate them at least generically. However, since separation from the other members of the family. we know nothing of the differences between their It is distinctive in possessing the long Sc, arched fore and hind wings, I have considered it advisable AIP1 + 2 and the relatively short 3A. The form of to group them into one genus until this important LIP is most peculiar, but it is interesting to note aspect of their structure is known. that a tendency for curvature of this vein occurs in Pseudogramma. An even more pronounced ap- Pleisiogramma, new genus proach to the condition of MP in Apsidoneura is Wings subpetiolate; Sc extending at least as far found in the Carboniferous Homaloneura similis as the fork of Rs; NIP either deeply forked or simple; Meunier, which resembles Jlexa in several other end of CuA and CUPweak distally and twigged; 1A respects, also. close to 2A proximally, more remote distally; 2A ending at about mid-wing or just before it. Pleisiogrammidae, new family Genotype: Pleisiogranzma medialis, n. sp. Wings moderately slender; costal area of moderate ~vidth;Sc long, extending beyond mid-wing; Rs Pleisiogramma medialis, n. sp. arising well before mid-wing, with 3 or 4 branches; Figure 11 JIP arising before origin of Rs, not strongly arched Length of wing, 17 mm.; width, 4.5 mm.; costal towards Rs; CuA and CUP straight and close to- margin straight or nearly so, with a slight curve gether, terminating just distad to mid-wing, much basally; apex rounded; posterior margin strongly shorter than in Paleuthygrammidae; CUP and 1A curved; Sc extending almost to very apex of wing; coalesced basally; only two anal veins well devel- Rs arising about one-fourth of wing length from oped, 3A reduced to a small vestige at base of wing; base, and forking into R2 + 3 and R4 + 5 just cross-veins numerous, but less so than in Paleuthy- beyond mid-wing; R2 + 3 forked soon after its grammidae. origin, and R2 also deeply forked; R3 and R4 + 5 This family is established for two new species, are slightly curved distally towards R2; MP arising both represented by isolated wings. Their venation slightly before the origin of Rs, unbranched; 2A

FIGURE11. P~~~S~O~T(II~IIII(I~t~f(/ifl/is,11 sp. Dritwing of holotypt'. IJetteri1lg as in figure 8. terminating well before mid-wing; SA very short only a small strip along the apical and posterior and close to posterior margin; cross-veins nulllerous, margin. The cross-reins are clearly preserved in the mostly sigmoidal. proximal part of the wing, but are obscure distally. Holotype: No. 4540ab, Museum of Comparative This species resembles the foregoing in the wing Zoology, in the upper layer of the Elmo limestone shape and the reduction of the third anal vein. It (F. M. Carpenter). The specimen consists of a very differs from medialis, however, in a number of well preserved wing, which might be either a fore or respects, such as the shorter Sc and forked media; hind wing. these differences and others may eventually require This species differs from all the other Caloneu- the erection of a distinct genus for reducta. Until a rodea, except Pseudogramma aberrans Mart., in complete specimen of either species (showing both having an unbranched MA. The possibility that fore and hind wings) has been found, and proven MA actually is forked and that its anterior branch the contrary, the possibility remains that medialis is coalesced proximally with Rs, does not seem to and reducta are based upon fore and hind wings of me at all likely. If this were the case, a portion of one species. However, in view of the nature of the MP leading to Rs should be visible. Since MP and differences between them, the chances of this seem Rs are strongly concave, this piece of MP would of very remote.

FIGURE 12. Pleisiogrccmtna reducia, n. sp. Drawing of holotype. Lettering as in figure 8. course also be concave; but all of the cross-veins Anomalogrammidae, new family between MP and Rs are strongly convex. Wings slender, but not subpetiolate; costal area of moderate width; Sc relatively short, terminating Pleisiogramma reducta, n. sp. at about mid-wing; space between Rl and costal Figure 12 margin, beyond end of Sc, very broad; Rs arising Length of wing, 15 mm.; width, 4 min. The wing well before mid-wing, forked; MP arising nearly at shape is essentially like that of medialis. Costal point of separation of Cu from R + M, deeply area a little broader than that of the latter; Sc forked; CuA not so close to CUP as in Paleuthy- terminating at about a third the wing-length from grammidae; CUP twigged distally; CUP and 1A the base, directly above the first fork of Rs; Rs coalesced basally; only two anal veins are present arising nearer the middle of the wing than in in the fossil on which tlle farnily is based, but a medialis and giving rise to 3 branches, R4 + 5 and third anal vein inigllt have been present in the R2 and R3; MP arising much nearer tlie base than original wing, since this part of the wing is poorly the point of origin of Rs and forked at about mid- preserved. Cross-veins convex but relatively few wing; its posterior branch continues the line formed in number. by the stem of MP, and its anterior branch is This family is erected for a species having a very directed gradually anteriorly; CUP and 1A close peculiar venation. Sc is shorter than in other mem- together at base, remote distally; 2A is somewhat bers of the order, Rl diverges away from the costal longer than in medialis; 3A is not preserved, but if margin beyond the end of Sc, and the cross-veins present it is obviously very short. are not at all numerous (i. e., for a Caloneurodean). Holotype: No. 4553ab, Museum of Comparative Because of these characteristics and several others Zoology, collected in the lower layer of the Elmo not occurring elsewhere in the Caloneurodea, I have limestone, 1927 (F. &I.Carpenter). This specimen considered a distinct family necessary for this insect. consists of a fairly well preserved wing, which lacks The family is apparently related to Paleuthygram- THE LOWER PERMIAN INSECTS OF KANSAS. PART 9 75

midae, but includes more highly specialized species far from mid-wing. Rl longer than in Paleuthy- with a greater reduction of the venation. grammidae, extending to the very apex of the wing; Rs arising before mid-wing, with 3 branches; MP Anomalogramma, new genus arising at point of separation of Cu from R, deeply Itrings with the proximal half much more slender forked; CuA and CUPclose together proximally, but than the distal half; Sc terminating at mid-wing; diverging distally; 1A remote frorn CUP;at least 2 JIP forked just below the end of Sc, its anterior anal veins present. branch arched slightly towards Rs at first, then This family was placed by Tillyard in the order curved away; CuA close to CUP proximally, but Neuroptera. He apparently overlooked the very diverging from it distally; CuA and CUP weak and close similarity between its venation and that of the twigged distally. Caloneuridae or Martynov's family Euthygram- Genotype: Anomalogramma parva, n. sp. midae, for neither of these is mentioned in his dis- cussion of the fossil. Martynov has already shown Anomalogramma parva, n. sp. that the true position of the Permobiellidae is with Figure 13 the Caloneurodea (1938a, 193810). The Permobiellid Wing: length, 9 mm.; width, 2 nini.; costal margin venation differs from that of all other families in the straight, apex rounded, ,posterior margin only order by the pronounced divergence of CuA and slightly curved; Sc close to R proximally but diverg- CUP distally as well as by the extraordinary con- ing abruptly towards costal margin soon after the vexity of the cross-veins, and the length of R1.

FIGURE 13. Ano~~~ulogra~t~~~~aparvcr, n. sp. Drawing of holotype. Lettering as in figure 8.

origin of Rs; Rs forked to about a third of its length; Genus Pe~wzobiellaTill. Rs arched towards MP just beyond mid-wing, then Permobiella Till., 1937, Amer. Journ. Sci., 33: 101; curved upwards towards R just before forking; CuA Martynov, 1938, 7: 80 slightly zigzagged distally, CUP more so, extending Wings with the proximal half much narrower than apically towards CuA along hind margin of wing; the distal half; Sc terminating very slightly beyond 1A nearly straight. The longer cross-veins are mid-wing; MP forked to about mid-wing, its ante- sigmoidally curved. rior branch very slightly arched towards Rs; CuA Holotype: No. 4561, Museum of Comparative slightly zigzagged distally and twigged; CUPparallel Zoology; collected in upper layer of Elmo limestone, to CuA proximally, but curved away distally. 1935 (F. M. Carpenter). This specimen consists of Cross-veins very strongly convex, more so than in a nearly complete bing, lacking only a very small any other genus of Caloneurodea. piece near the apex. The preservation is good for Genotype: Perrnobiella perspicua Till. such a small wing, though many of the cross-veins are not discernible. Whether the specimen is a fore Pernzobiella perspicua Till. or hind wing cannot be determined at present. Figure 14 This insect is the smallest member of the Calo- neurodea so far known. The venation as a whole is Permobiella perspicua Till., 1937, Amer. Journ. Sci., fundamentally like that of the Paleuthygrammidae 33: 102 but differs in the structure of the individual veins, Wing: length of wing fragment (holotype), 11 as indicated under the genus. inln.; width, 3.6 mm.; the basal part of the wing, proximal to the origin of MI', is missing; the com- Family PERM~BIELLIDAE plete wing was probably about 14 mm. long. Costal Wings moderately slender, not subpetiolate; nlargin slightly arched, apex probably pointed (not costal area of moderate width; Sc terminating not rounded as drawn by Tillyard); posterior border nearly straight; R4 + 5 is ahout one-half as long as conspicuous than the longitudinal veins.g There is the entire Rs; R2 and R3 are nearly as long as a noticeable difference in the number and arrange- R4 + 5; between the branches of Rs, tlie cross-veins ment of the cross-veins between the holotype of form short intercalated veins; nearly all cross-veins perspicua and the specimen in the Harvard collec- in the wing are sigmoidally curved and are dark tion. This discrepancy is almost certainly due to brown. individual variation, rather than to specific differ- Holotype : No. 15593, Peabody Museum. This ence. The occurrence of several incomplete cross- consists of a nearly complete wing, lacking the veins in the Harvard specimens is strong support of proximal fifth. this conclusion. In the Museum of Comparative Zoology there is one specimen (no. 4552ab, lower layer of the Elmo Order PR.OTORTHOPTERA limestone) ; this consists of the distal half of a wing. This order was established by Handlirsch in 1906 Tillyard's figure of perspicua is incorrect in several for a series of Carhoniferous and Permian insects, respects. The apex of the wing is partially obscured presumably transitional between the Paleodictyop- by a plant fragment in the holotype, and the part tera and the (s. 1.). Since its erection, of the apex that is preserved is obviously distorted. the order has become a vague assemblage of all Paleozoic insects which show traces of Orthopteroid characteristics. The taxonomic confusion of these fossils has been increased by the gradual diminution of the distinctions between the Protorthoptera and Protoblattoidea, another Paleozoic order erected by Handlirsch for species supposedly intermediate be- tween the Blattaria and Paleodictyoptera. In recent years, an extensive series of Carboniferous and especially Permian species has been described which cannot be placed definitely in either of these orders. Many attempts have been made to divide the order into several suborders (e. g. Handlirsch, 1937), but there have always remained a great many families and genera which failed to fit into the pro- posed divisions. Several of the suborders thus formed have subsequently been completely removed from the Protorthoptera and set up as distinct orders (e. g. Caloneurodea), but these moves have FIGURE14. Pe-rntobiella perapicua Till. A, drawing had little effect on the extreme heterogeneity of the of holotype; B, drawinp of specimen no. 4552, Mus. remaining Protorthoptera. Martynov has at- Comp. Zool. Lettering as in figure 8. tempted (193813) to obtain some homogeneity by restricting the order so as to include only the Paleo- zoic saltatorial Orthopteroids; the non-saltatorial The apex is perfectly preserved in the Harvard forms he places in a distinct order, Paraplecoptera. specimen, however, and is clearly much more pointed Even if such a classification is assumed to have a than it is represented in Tillyard's drawing. R1 is phylogenetic basis, which seems dubious, it is obvi- not strongly curved distally, as he drew it, but terminates before the apex of the wing. IIe was This is indicated clearly in the photograph of tlie also under the erroneous impression that the holo- Harvard specimen of P. perspicua published in the type was practically complete and he indicated the Annals of the Carnegie Museum, 1934, vol. 22, pl. 24, base of the wing just beyond the limits of the fossil. fig. 2. It should be noted that as a result of an error In view of the structure of other Caloneurodean which escaped my attention in the proof, the caption wings, however, I believe this wing was much longer under this figure is incorrect. It should read, "A Caloneurid from the Lower Permian of Kansas, No. than he supposed. 4552, Museum of Comparative Zoology. Note the Tillyard noted, and also indicated by the specific thickness and convexity of the cross-veins. About name perspicua, that the cross-veins of this wing x5." This photograph of perspicua was included for were remarkably convex. Their dark brown color comparison with Calorteurella carbonaria, described in adds to their prominence and makes them far more tlie text. THE LOWER PERMIAN INSECTS OF KANSAS. PART 9 77 ouslg impractical, for the leg structure is unknown in species in this family, but Tillyard, after a study of more than four-fifths of the species which have been nearly a hundred specimens in the Yale collection, placed in the Protorthoptera. was able to demonstrate that all of these genera and In my opinion, most of the difficulties encountered species really represented a single variable species, in the of the Protorthoptera have resulted Probnis speciosa Sellards.ll From my study of the from two factors. In the first place, many of the Harvard collection of Probnisids, numbering about species, genera and families have been founded on 250 specimens, I am convinced that his synonymy small ~ingfragments, too incomplete to permit is absolutely correct and also that the family Prob- determination of their affinities. Such fragmentary nisidae is now known to us by only this species. material will have to be ignored in the formation The representatives of this insect in Sellards' coI- of a taxonomic system within the order. In the lection consisted of isolated fore and hind wings, second place, many of the species have been so but a number of the specimens which were in the incompletely and inaccurately described that their Yale collection and which were described by Tillyard true affinities have not been apparent. The majority show parts of the body. In the Harvard collection of the known Protorthoptera are from the Carboni- there are many additional specimens with body ferous and have been described by Scudder, Brongni- structures preserved, so that we are now able to art, Meunier and Handlirsch. The figures and obtain an unusually good concept of the general descriptions of Scudder and Brongniart, though structure of this insect. Sellards placed the Probni- accurate and complete enough for the need of their sidae in the Order Protorthoptera, but Tillyard times, do not include many details (as convexity and concluded from his study of the Yale fossils that concavity of the veins) which are now essential; they were aquatic in their nymphal stages and there- those of Meunier, as is well known, are totally unre- fore belonged to the Order Protoperlaria. I do not liable; and those of Handlirsch, though better than agree with this conclusion. I believe that the Lleunier's, are usually incomplete and very erratic.1° nymphs of Probnis were terrestrial and that the The deplorable state of the literature on the Protor- genus was closely related to the Geraridae, Ischno- thoptera convinced me several years ago that a neuridae and other Orthopteroids which have been "revision" of the Carboniferous Protorthoptera, placed in the Protorthoptera. The reasons for this based on published figures and descriptions, would view, as well as corrections to Tillyard's comments in no way ameliorate the taxonomy of these insects. on my concept of the Protoperlaria, will follow the For this reason I have spent a large part of the past detailed description of Probnis speciosa. three years in a study of the type specimens of this The Probnisidae may now be characterized as group of fossils, both European and North American. follows: Insects of moderate size. Head rather This task has now been completed and the drawings small, eyes prominent and protruding laterally; and descriptions are being assembled for publication. antennae shorter than the thorax, robust. Thorax A discussion of the classification of the Protorthop- large and broad; prothorax short, with a small tera will be included there. In the present paper and notum and a pair of moderate-sized but distinct in the subsequent part in this series, I shall use the lateral lobes, slightly overlapping the mesothorax; term Protorthoptera in a broad sense, and shall not legs heteronomous, the prothoracic pair much attempt to anticipate the account of the Carboni- shorter than the hind or middle pair; the tibia of ferous species. all legs forming distally a series of conspicuous finger-like projections, these being largest on the Family PROBNISIDAE fore legs; tarsi apparently 3-segmented. Abdomen The family Probnisidae was established by Sel- long and robust; cerci short. Fore wing: slightly lards (1909) in the Protorthoptera for a series of granular in appearance, coriaceous, but not elytri- insects from the Elmo limestone having coriaceous form; Sc terminating on the costa before the apex fore wings and a Protorthopterous venation. Up to of wing; R1 unbranched, extending to apex; Rs the present time the family has not been found in arising in the proximal half of the wing, unbranched; any other formation, so that our knowledge of it is MA and MP separating at about the level of the restricted to the data provided by the Kansan fos- origin of Rs, both of these veins very variable in sils. Sellards described four genera and eight respect to branching; CuA and CUPseparating close to the base of the wing; CuA dividing shortly after lo For an example of the confusion resulting from the its origin into two main branches, CuAl and CuA2; work of Handlirsch and Meunier, see my note on the Commentry Caloneuridae (Carpenter, 1934, p. 325- " With the possible exception of one unrecognizable 326). species, Lecopterum delicosum Sell. 78 CARPENTER

CuAl gives rise to a variable number of arched the venation under the species. In all probability branches along the posterior margin of the wing; such general body characteristics as the relative CuA2 usually simple; CUP unbranched, straight; size of the prothoracic lobes and the relative length 1A straight and usually simple. Hind wing: mem- of legs and cerci are of generic rank; and very likely praneous, but with a definite trace of the wrinkles also such general venational characteristics as the which are better developed in the fore wing; costal approximate position of the origin of Rs and MA space much narrower than in the fore wing; Sc and are in the same category. I regard the peculiar R1 essentially as in the fore wing; Rs arising almost coriaceous texture of the wings as a family trait, at the very base of the wing, unbranched; MA and rather than a generic one, as it was treated by MP simple (usually) or forked; CuA separating Tillyard. from CUP at the very base of the wing and giving Genotype: Probnis speciosa Sellards. rise to several arched branches similar to those of the fore wing, except that they are usually longer; Probnis speciosa Sellards CUP straight and unbranched, terminating at the Figures 15, 16 re-entrant angle of posterior margin; anal fan well Probnis speciosa Sellards, 1909, Amer. Journ. Sci., developed, with at least six main veins, all of these 27: 159; Tillyard, 1937, ibid., 33: 417 arising from a single proximal stem. Probnis coriacea Sellards, 1909, ibid.: 160 Tillyard recognized two genera in the family in Espira obscura Sellards, 1909, ibid.: 160 addition to Probnis itself: Lecopterum Sellards and Stoichus elegans Sellards, 1909, ibid.: 160 Telactinopteryx Tillyard. The former genus was Stoichus arcuatus Sellards, 1909, ibid. : 161 based upon a single fore wing (delicosum Sell.), Stoichus minor Sellards, 1909, ibid.: 161 which was inadequately described by Sellards, with- Stoichus tenuis Sellards, 1909, ibid.: 161 out a figure. The type specimen was missing from Stinus brevi-cubitalis Sellards, 1909, ibid. : 161 the Sellards' collection when I studied his types in Probnis sellardsi Handlirsch, 1919, Denkschr. der 1927. Since the identification of further specimens Akad. Wiss. Wien, 96: 33 of delicosum is almost hopeless, as pointed out by Fore wing: length, 9-14 mm.; width, 34mm. Tillyard, I propose to regard the name Lecopterum Costal space narrow, almost always with a few vein- delicosum as a nomen nudum. Telactinopteryx was lets, sometimes with many; Sc ending at about based upon a single species (striatipennis Till.) and mid-wing, its actual point of termination usually placed in a separate subfamily. In my opinion this obscure because of the vein's faintness; Sc is nearly species does not belong to the Probnisidae. Its eyes straight in most specimens, but is distinctly curved are very small, the prothoracic lobes almost absent, in others; R usually joined to Sc or to the costal the tibiae are unmodified, the cerci multisegmented, margin by several veinlets, which are oblique and the wings lack the granulation described above in often forked beyond the end of Sc; R1 and Rs Probnis, and their venation is distinctly different unbranched; MA and MP consistently diverging at from that of the latter genus. I therefore establish about the level of the origin of Rs; both MA and a distinct family, Telactinopterygidae, for this MP are unbranched in about 50Yo of the specimens, genus. Probnis cbnsequently remains the only but either one or both may be forked, though the known genus of the family Probnisidae. forks do not extend at most to more than a third of the wing length from the apex; MP is usually weak Genus Probnis Sellards proximally; CuA is the strongest vein in the wing Probnis Sellards, 1909, Amer. Journ. Sci., 27: 159; and it is also the most variable; CuA2 usually arises Tillyard 1937, ibid., 33: 417 quite near the base of the wing. The marginal, Espira Sellards, 1909, ibid., 27: 160 arched branches of CuAl vary from four to nine, the Stoichus Sellards, 1909, ibid., 160 distal branches being longer than the proximal ones; Stinus Sellards, 1909, ibid., 161 CUPis always weak and straight; the anal veins are Since only one species of this genus is known, and highly variable, but at least 1A and 2A are usually that one species has a very variable venation, it is recognizable. impossible to do more than to suggest generic Hind wing: shorter than fore wing, 7-12 mm. in characteristics. Tillyard has given a very detailed length; Sc close to margin, ending beyond mid-wing, account of the venation under the genus, but nearly usually with a series of short distal veinlets; R1 all of his statements are clearly specific rather than with longer distal veinlets between it and the costal generic in nature; Tillyard himself apparently real- margin; Rs arising from R1 at a point nearer the ized this, since he completely omitted description of base of the wing than in the fore wing; Rs always THE 1,OWER PERMIAN INSECTS OF KANSAS. PART F)

FIGURE15. P~obnisspen'osa Sell. Restoration based on specimens in t,he Harvard collection. Lettering as in figure 8. unbranched (not forked as drawn by Tillyard); MA longitudinal veins in the wing proper are distinct and MP separating at a point nearer the base than and their structure is essentially like that shown in in the fore wing; both may be either forked or the accompanying figure (15). All other specimens simple; CuA with at least two distal branches, usu- of hind wings of this insect, whether isolated or ally three; CUP straight; 1A very nearly straight; attached to fore wings, have a similar venation. all the other anal veins may be an extension of 2A, Many specimens in the Harvard collection show since all arise from a single stem; the first vein below various body structures. The following description 1A is simple, the next forked and the others usually and measurements are based upon this material. simple. The entire body, exclusive of the cerci, is about as It should be noted that Tillyard's description and long as the fore wing. The head, which is apparently drawing of the hind wing of speciosa are incorrect prognathous, is about as wide across the eyes as it in several respects. Rs arises much nearer the base is long (1.8 mm.). The eyes are strongly protu- of the wing than is shown in his figure and it does berant, as they were depicted by Tillyard in his not fork; M is forked at about the level of the point restoration of the head (1937, figure 7). The anten- where he has indicated the origin of Rs, instead of nae, which are approximately 2 mm. long, are at mid-wing; CuA has two distal branches; between inserted just in front of the eyes and contain about 1A and the vein which he labels 2A there is a simple 19 segments. The thorax is large and massive. straight vein. His drawing was based on an isolated The prothorax is small (.9 mm. long) and the pro- hind wing in the Yale collectiorl (15560); I have thoracic lobes are of moderate size (I mm. long, examined this specimen very carefully and compared 2 mm. wide), and slightly overlap the mesothorax. it with hind wings in the Harvard collection. The The mesothorax is the largest thoracic segment specimen figured by him is very poorly preserved, (1.5 mm. long, 2.3 mm. wide); the mesonotum is and the anal veins connot be distinguished satis- large and more or less cordate in form. The meta- factorily. Under oblique illumination, however, the thorax is smaller than the preceding segment (1.2 80 CARPENTER mm. long, 2 mm. wide) and has a smaller, subcordate the Harvard collection. All structures shown notum. The three pairs of legs are markedly dif- in the figure are preserved in at least one of the ferent in size. The fore pair are 3 mm. long from fossils, except the middle and fore tarsi, which have the base of the femur, the middle pair, 3.8 mm. long, been drawn like the hind pair. and the hind pair 5.5 mm. long. The femur of all There are two features of this species which are legs is moderately robust and of ordinary form, but especially noteworthy. One of these is the texture the tibiae are most peculiarly modified. Each tibia of the wing membrane and the other is the form of terminates distally in a group of four finger-like the tibiae. As both Sellards and Tillyard stated, projections.12 The fore tibiae, which are distinctly the fore wing has a distinctly coriaceous appearance. flattened and broadened distally, bear the longest In all probability, however, the wing was not as projections. The processes are obviously not move- thickened and heavy as that of even most cock- able spines but appear to be rigid extensions of the roaches (such as Blaberus), for a great many of the tibia itself. Several large, oblique spines are present specimens of Probnis have the fore wings twisted or along the sides of the hind tibiae, though I have not folded, a condition which would hardly have taken seen them on the other legs. The tarsus is attached place had the wing been very stiff. As Tillyard has to the tibia between two of the projections, appar- mentioned, the granular appearance of the wing is ently slightly on the dorsal surface of the tibia. All produced by the presence of protuberant bases of tarsi are small (about .7 mm.); those of the hind legs are 3aegmented, the first two segments being small, and the third about three times as long as the others.l3 I have not been able to determine the nature of the segmentation of the middle and fore tarsi but it was almost certainlv like that of the hind tarsi. The abdomen is nearly homonomously segmented. There is absolutely no indication of lateral gill-vestiges, like those in the Lemmato- phoridae (Protoperlaria). The cerci are stout and composed of 7 segments. Types: the holotype of speciosa, originally in the Sellards' collection, was lost in 1927. A neotype (no. 15518) was selected by Tillyard from the Yale material. FIGURE16. P~obnisspeciosa Sell. A, Drawing of This insect is a common one in the Elmo lime- fore tibia, based on specimen no. 4615 and 4613, Mus. stone. Tillyard listed 99 specimens in the Yale Comp. 2001. B, drawing of hind tarsus and part of collection; I find 37 in the Sellards' collection and tibia, based on specimen no. 4619, Mus. Comp. Zool. about 250 in that of the Museum of Comparative Zoology. These make a total of about 385 speci- microtrichia, which are almost evenly distributed mens. The accompanying restoration of this insect over the surface. In only one specimen have I (figure 15) is based upon several specimens in actually seen any of the microtrichia themselves; in this fossil they are very short and thin. Tillyard l2 Tillyard's figure of the "fore leg" of speciosa (1937, fig. 7) does not indicate these tibia1 projections. has stated his belief that the fore wings had soft, My examination of the specimen on which he based wavy hairs, forming a woolly covering. So far as I his drawing (no. 15548, Peabody Museum) has shown can determine, however, he did not actually see that the leg figured was actually one of the middle pair such hairs and I have failed to discern them in any and that the projections were broken away in the specimens, including those in the Yale collection. counterpart of the fossil, which was not collected. Oddly enough, the subcosta of the fore wing was One fore leg of this specimen was preserved but it was armed with a series of large spines, on the under- covered by limestone matrix and was consequently not surface only. These spines, which are directed pos- seen by Tillyard. I have subsequently removed the teriorly extend almost the entire length of the vein. matrix, so that the complete tibia, with the projections described above, is now visible. The significance of the wing texture and of the l3 Tillyard has drawn the tarsus of speciosa (1937, presence of the spines on the subcosta are not at all fig. 7) with 5 segments, but this segmentation was pure apparent; they may be in some way associated with assumption on his part, since, as he states, none of the the insect's habits or with the environment in which Yale specimens showed the tarsal segmentation. it occurred. In connection with the modified tibiae, THE IAOWER PERhlIAN INSECTS OF KANSAS. PART 9 81 it is noteworthy that the tarsi are distinctly sup- phoridae (the family 011 wliich the Order Protoper- pressed. I have the distinct impression that the laria was based) were likewise present in at least tarsi arose from the dorsal surface of the tibiae some of the Protorthoptera and could not therefore rather from the ventral side. The function of the be used to distinguish these two orders, as had been tibia1 projections is very puzzling. They hardly proposed by Tillyard. seein of predatory significance, though this would be 2. The wing venation of the Lemmatophoridae borne out by the prognathous head. The fore legs was so similar to that of some of the Protorthoptera resemble those of the Gryllotalpids, and certainly that this alone could not distinguish the two groups. suggest a fossorial function, but the structure of the 3. The nymphs of the Lemmatophoridae were other legs and of the rest of the insect does not sup- aquatic, possessing abdominal tracheal gills and port this indication. At present I am inclined to swimming legs. (This statement was based upon the belief that the tibiae might have played a part an extensive series of nymphs from the Elmo in climbing; this is consistent with the presence of ' limestone in the Museum of Comparative Zoology.) the strong spines on the hind femora and of the 4. The Lemmatophoridae lacked a fully developed modification of all tibiae. external ovipositor. In synonymizing seven of Sellards species and 5. Most, if not all, of the families of Protorthop- three of his genera with speciosa, Tillyard was tera in the Elmo limestone included species with a unquestionibly correct, and I have considered it fully developed external ovipositor in the female. unnecessary to present further evidence for this My final conclusions were that the Order Pro- from the material in the Harvard collection. toperlaria should be restricted to include only species Handlirsch established one additional species for an having aquatic nymphs; and that so far as known, incomplete hind wing which Sellards figured without the Lemmatophoridae was the only family that attempting to determine its generic position. This would fit into the order as so defined. is of course another synonym. My views on the relationship between the Pro- Nothing is known about the immature stages of toperlaria and Protorthoptera were stated as follows: Probnis. Tillyard has stated that the abdomen of the order Protoperlaria "stands in direct evolution- the adult shows a mottled pattern which assisted ary line leading to the Perlaria, although it is very him to recognize as belonging to this species one of unlikely that any of the genera now known were the the commoner larval types in the Yale c~llection.~~progenitors. The fossils indicate that since the His statement regarding the mottled pattern of the Lower Permian the evolution of this Perlarian line adult abdomen puzzles me greatly, for I have seen has resulted in the reduction of the number of tarsal no indications of such a pattern in any specimens, segments, the elimination of the prothoracic lobes including those in the Yale collection. I do not be- and the posterior media of the fore wing, the nearly lieve there is any evidence whatever for associating complete loss of the vestiges of an external ovipositor the nymphs mentioned with the genus Probnis. and the lateral abdominal gills of the nymphs and adults. In respect to these and other features the RELATIONSHIPOF THE PROBNISIDAE Protoperlaria are intermediate between the true ,4fter his study of the Yale specimens Tillyard Protorthoptera and Perlaria, for 5-segmented tarsi, came to the conclusion that the Probnisidae were prothoracic expansions, an external ovipositor were aquatic in the nymphal stages and belonged to the possessed by at least some of the Protorthoptera. Order Protoperlaria, and he presented a series of The close relationship between the Orthoptera (s. arguments to support this view. I do not believe lat.) and the Perlaria is now generally recognized by that he has really proved his case, however. It is entomologists, and the Perlaria are frequently my conviction that the Probnisidae had terrestrial regarded as Orthoptera which adopted an aquatic nymphs and that they were Protorthopterous insects mode of existence in the immature stages, forming related to the Geraroidea. Before discussing the a divergent branch of the Orthoptera. The Proto- evidence bearing on this question, however, we must perlaria of the Lower Permian appear to represent first consider the differences between the Protoper- the beginnings of that divergence."'" laria and Protorthoptera. In my account of the Protoperlaria of the Elmo limestone (Part 7 of this Zeuner (1939, 1). 47) has objected to the it~clusion series), I showed that: of insects with prothoracic expansions in the Protor- 1. Prothoracic lobes like those of the Lemmato- thoptera on the grounds that "Handlirsch defined the Protorthoptera by the absence of these expansions." l4 'l'hroughout his paper Tillyarcl uses the tern1 lurrirtc I am afraid that he is mistaken in this, for Handlirsch in a broad sense, i. e., including nymphs. makes no mention of either tlie absence or presence of In spite of these statements, wllicll I corisicler to I have quoted this passage in full in order to pre- be clear, Tillyard nlisconstrued nly remarks in his sent in Tillyard's own words the evidence for his discussion of the Probnisidae, implying that I dis- conclusion. Briefly stated, of ,course, he claims to tinguished between the Protoperlaria and Protor- have found specimens of aquatic nymphs in the thoptera only by the presence or absence of the Yale collection which belong to Probnis. He gives ovipositor, and that I attempted to prove a correla- two reasons for associating these nymphs with the tion between the absence of a com~leteexternal adults of the latter genus:- I) Since Probnis is one of ovipositor and the aquatic habitat of the nymphs. the most abundant insects represented in the Yale He then comes to the conclusion, which he appears collection, its nymph, if aquatic, ought to be fairly to regard as original with him, that the Protoper- common. 2) One of the "common" nymphs with laria should be restricted to include "those forms aquatic niodifications has an abdominal pattern like which possessed aquatic larvae [nymphs.]" This is that of the adults of Probnis. precisely the conclusion which I had previously In my opinion, Tillyard was not justified in reached. drawing the conclusions which he reached. Follow- Having adopted this restriction for the Protoper- ing Tillyard's death, in 1938,the Yale specimens from laria, Tillyard places the Probnisidae in that order. the Elmo limestone were turned over to me for study His reasons for doing so are given by him as follows: by the authorities of the Peabody Museum. In- "Now there are nineteen specimens of larvae cluded in a box marked "larval types" are the [nymphs] in the Yale collection. Of these one be- nymphs to which Tillyard obviously had reference longs to a large Protorthopteron . . . The other in his paper on the Probnisidae. I find in this box eighteen are all evidently aquatic types, either com- twenty-one nymphs (two more having turned up in plete or nearly so, and they all belong quite as the collection since Tillyard wrote his account of evidently to the Protoperlaria, and not to Plec- Probnis), of which fifteen are like those described by toptera or Odonata, though both of these Orders me as Lemmatophorids, and which obviously belong also occur in the Kansas beds. to that group, as stated by Tillyard. The other six "While retaining completed results of my larval specimens comprise the Protorthopteron mentioned study for a later paper, it is necessary to state here by Tillyard and five poorly preserved nymphs which that I have been able to recognize two quite distinct are quite different from the Lemmatophorids. types of larvae in the Yale collection over and above These are without doubt the specimens which Till- those described by Carpenter, and that I feel confi- yard regarded as the young of Probnis. Similar dent that these two types represent the larvae of the ' nymphs are present in the Harvard collection, but genera Paraprisca (Lemmatophoridae) and Probnis I did not mention them in my account of the imma- respectively. As Probnis is one of the most abundant ture stages of the Lemmatophoridae, because I felt types represented in the collection, we should expect certain that they did not belong there. All of these to find its larva fairly common also, if it were an nymphs are characterized by having a pigmented aquatic types; and such, indeed proves to be the case. pattern on the dorsum of the thorax and abdomen. "We now have a prima facie case for including the Tillyard's statement that these nymphs are "fairly Probnisidae within the Protoperlaria. They fulfill common" in the Yale collection is very surprising the definition laid down by me [sic!] earlier in this in view of the actual presence of only five specimens. paper that they should possess aquatic larvae and Even in the combined Harvard, Yale and Sellards prothoracic expansions or lobes in the adult." collections from Elmo, an aggregation of over 10,000 specimens, I find a total of but 15 of these nymphs. such expansions in his definition of the order (Die They appear to me to be very rare rather than fairly' fossilen Insekten, 1906, p. 128). Handlirsch's statement common, and as such would hardly fit Tillyard's about the prothorax is: "der Prothorax gross, oft stark requirements for the nymphs of Probnis, viz., "As verlangert. " However, even if it were true that Hand- Probnis is one of the most abundant types repre- lirsch's definition of the order characterized the Protor- sented in the collection, we should expect to find its thoptera by the absence of prothoracic expansions, that larva also fairly common if it were an aquatic type." would not prevent the inclusion of species with them in As to the pattern on the abdomen and thorax, I the order. The concept of any taxonomic category, whether order, family, genus, etc., is continually being have been completely unable, as stated above, to changed as new species are found. A flexible concept find any trace of this on the adults of Probnis, is especially necessary for extinct, categories, since our including the material in the Yale collection. There knowledge of them is acquired only by the accumulation are, of course, the irregular blackened areas caused of a few facts at infrequent intervals. by the breaking away of part of the carbonized body; TIIE LOWER PEltMIllN INSECTS OF KANSAS. PAItT Y 83 but such areas are nearly always present in all the distinguish ordinal differences on the wing venation Elmo insects which have body parts preserved. It alone. is possible that Tillyard misinterpreted the signifi- It is my conviction, therefore, that the Probni- cance of these blackened areas, although one would sidae had terrestrial nymphs and were members of suppose that he would have noted them in the Lem- the Order Protorthoptera (suborder Geraroidea). lnatophoridae and other common species. Of course, Their closest relatives as far as known were the if the adult Probnisidae actually had a mottled Lepiidae, a family represented in the Elmo limestone pattern on the abdomen even remotely resembling by Lepium elongatu~~zSellards. This latter insect, that of the nymphs mentioned, I would consider which was referred to the Protorthoptera by Till- this strong and even conclusive evidence of the yard, will be dealt with in the next part of this series association between the nymphs and the adults. of publications. There is one structural feature of the Probnisid adults which indicates that their immature stages REFERENCES were not aquatic. The adults of the Lemmato- CARPENTER,F. M. phoridae (Protoperlaria) possess lateral abdominal 1931a. The Lower Permian Insects of Kansas. projections which have been interpreted by Tillyard Part 3. The Protohymenoptera. Psyche, and me (Carpenter, 1935, p. 143) as vestiges of the 37: 343-374. lateral tracheal gills present in the nymphs. No 1931b. Same, part 4. The Order , and Additions to the and such structures occur in the adults of Probnis, as Protohymenoptera. Amer. Journ. Sci., (5) Tvas observed by Tillyard. Kow while it is of course 22: 113-130. true that the presence of such gill vestiges is not at 1933. Same, part 6. Delopteridae, Protelytrop- all necessary in adults having aquatic nymphs, tera, Plectoptera, and a new collection of nevertheless their constant presence in all known Protodonata, Odonata, Megasecoptera, genera of Protoperlaria would at least lead us to Homowtera and Psocowtera. Proc. Amer. expect them (but not require them) to be present in Acarl. Arts Sci., 68: 411-503. other members of the group. Their absence in the 1934. Carboniferous Insects from Pennsylvania adult Probnisidae is at least an indication, however in the Carnegie Museum and the Museum slim, that these insects did not have aquatic nymphs. of Comparative Zoology. Ann. Carnegie Mus., 22: 323-341. If the nymphs of the Probnisidae were terrestrial, 1935. The Lower Permian Insects of Kansas. then the family is eliminated from the Protoperlaria, Part 7. The Order. Protoperlaria. Proc. as tliat order has been defined. In this connection, Amer. Acad. Arts Sci., 70: 103-146. it is pertinent to note that a comparison of the 1936. Revision of the Nearctic Raphidiodea structure of the Probnisidae (adults) with that of (Fossil and Recent). Proc. Am~r.Acad. the Lemmatophoridae, on which the order Proto- Arts Sci., 71: 89-157. perlaria was established, gives added support to the 1939. The Lower Permian Insects of Kansas. view that the Probnisidae are not so closely related Part 8. Additional Megasecoptera, Pro- to the Lemmatophoridae as Tillyard believed. The todonata, Odonata, Homoptera, Psocop- head of the Probnisids was prognathous, whereas tera, , Plectoptera, and Protoperlaria. Proc. Amer. Acad. Arts that of the Lemmatophorids was liypognathous; the Sci., 73: 29-70. antennae of the Probnisids were relatively short and 1940. Revision of the Nearctic Hemerobiidae, stout, whereas those of the Lemmatophorids were Berothidae, Sisyridae, Polystoechotidae very long, slender, and multisegmented; the tarsi of and (Neuroptera). Proc. Amer. the Probnisidae were 3-segmented and the tibiae Acad. Arts Sci., 74: 193-280. were peculiarly modified, whereas the tarsi of the COMSTOCK,J. H. Leinmatophorids were five segmented and the tibiae 1918. The Wings of Insects. Ithaca. were normal; the cerci of the Probnisidae were short HANDLIRSCH,A. and had few segments, whereas those of the Lemma- 1906-1908. Die fossilen Insebten. Wien. Denk- tophorids were long and multisegmented. The vena- schr. der Akad. Wiss. Wien, 96: 1-82. 1919. Revision der palaozoischen Insekten. tion of the Probnisidae is like that of the Lemma- 1937. Neue Untersuchungen uber die fossilen tophorids in most respects, it is true; but I have Insekten. 1. Ann. Natnrhist. Mus. Wien, repeatedly pointed out that the venation of the 48: 1-140. Protoperlaria, as represented by the Lemmato- R~ARTYNOV,A. V. pliorids, and that of inany of the Perillinn Protor- 1925. To the k~iowledgeof fossil insects from tlioptera are so much alike tliat it is not possible to Jurassic beds in Turkestan. 2. Raphidiop- tera, Ortlloptera (s. l.), Oclonata, Ncurop- t~lattinidae and Protorthoptera. Amer. tera. Bull. Acad. Sci. Russie, 1926: 569- Journ. Sci., (4) 27: 151-173. 598. TILLYARD,R. J. 1928. Permian fossil insects of North-East Eu- 1925. Kansas Permian Insects. Part 5. The rope. Trav. MusEe G6o1. prb l'Acad. Sci. Orders Protodonata and Odonata. Amer. URSS. 4: 1-118. Journ. Sci. (5) 10: 41-73. 1930. On two new Orthopteroid insects from the 1926. Upper Permian Insects of New South Permian deposits of the Gov. of Perm. Wales. Part 2. The Orders Mecoptera, Ann. Soc. Pal. de Russie, 8: 3547. Paramecoptera, and Neuroptera. Proc. 1932. Permian fossil insects from the Arkhangelsk Linn. Soc. New South Wales, 61: 265-282. district. Part 11. Neuroptera, Megalop- 1932. Kansas Permian Insects. Part 14. The tera, and Coleoptera, with the description Order Neuroptera. Amer. Journ. Sci., (5) of two new from Tikhie Gory. 23: 1-30. Trav. de 1'Inst. PalBoz., 2: 63-96. 1936. Same, Part 16. The Order Plectoptera 1938a. Permian fossil insects from the Arkhan- (cont'd.): The family Doteridae, with a gelsk District. Part V. The family Euthy- Note on the Affinities of the Order Proto- grammidae and its relationships (with the hymenoptera. Amer. Journ. Sci., (5) 32: description of a new genus and family from 435-453. Chekarda). ~rav.de 1'Inst. PalBont., 7(3): 1937. Same, Part 17. The Order Megasecoptera 69-80. and Additions to the Palaeodictyoptera, 1938b. Etudes sur l'histoire ghlogique et de Odonata, Protoperlaria, Copeognatha, and phylogknie des ordres des insectes (Ptery- Neuroptera. Amer. Journ. Sci., (5) 33: gota). 1. et -Poly- 81-110. neoptera. Trav. de 1'Inst. PalBont., 1937. Same, Part 19. The Order Protoperlaria 7(4): 1-149. (cont'd.). The Family Probnisidae. Amer. SELLARDS,E. H. Journ. Sci., (5) 33: 401425. 1906. Types of Permian Insects. Amer. Journ. ZEUNER,F. E. Sci., (4) 22: 249-258. 1939. Fossil Orthoptera Ensifera. British Mus. 1909. Same. Part 3. Megasecoptera, Orycto- (N. Hist.). London.

EXPLANATIONOF PLATE1. FIGURE1. Megatypus schucherti Till. (xl) Photo- graph of specimen no. 4633, Museum of Comparative Zoology. FIGURE2. Apsidoneura Jiexa, n. sp. (x6). Photo- graph of holotype, no. 4572, Museum of Comparative Zoology.

Bibliography of the

Bibliography of the Neuropterida Reference number (r#): 1853

Reference Citation: Carpenter, F. M. 1943 [1943.??.??]. The Lower Permian insects of Kansas. Part 9. The orders Neuroptera, Raphidiodea, Caloneurodea and Protorthoptera (Probnisidae), with additional Protodonata and Megasecoptera. Proceedings of the American Academy of Arts and Sciences 75:55-84.

Copyrights: Any/all applicable copyrights reside with, and are reserved by, the publisher(s), the author(s) and/or other entities as allowed by law. No copyrights belong to the Bibliography of the Neuropterida. Work made available through the Bibliography of the Neuropterida with permission(s) obtained, or with copyrights believed to be expired.

Notes:

File: File produced for the Bibliography of the Neuropterida (BotN) component of the Global Lacewing Digital Library (GLDL) Project, 2007.